A Randomised Clinical Trial of Dialectical Behaviour Therapy and Conversational Model for the Treatment of Borderline Personality Disorder: A Hybrid Efficacy- Effectiveness Study in a Public Sector Mental Health Service in Australia
Dr Carla Joy Walton BSc (Psyc) (University of New South Wales) DPsyc (Clin) (University of Wollongong)
Thesis submitted for the degree of Doctor of Philosophy January 2018
School of Medicine and Public Health University of Newcastle, Australia
This research was supported by an Australian Government Research Training Program (RTP) Scholarship
Statement of Originality
This thesis contains no material which has been accepted for the award of any other degree or diploma in any university or other tertiary institution and, to the best of my knowledge and belief, contains no material previously published or written by another person, except where due reference has been made in the text. I give consent to the final version of my thesis being made available worldwide when deposited in the University’s Digital Repository **, subject to the provisions of the Copyright Act 1968. **Unless an Embargo has been approved for a determined period. Statement of Collaboration I hereby certify that the work embodied in this thesis has been done in collaboration with other researchers. I have included as part of the thesis a statement clearly outlining the extent of collaboration, with whom, and under what auspices. The PhD candidate designed the study in consultation with the PhD supervisors, prepared and submitted the ethics application, and maintained ethical reporting requirements. The PhD candidate established agreement with the local health district for the study to be conducted; set up processes and documentation for recruitment; supervised research assistant staff; and managed the research on a day-to-day basis. The PhD candidate monitored data collection, scored and cleaned the data and conducted statistical analyses, with support of the local health district Research Manager. The PhD candidate co-supervised two postgraduate clinical psychology research projects. These are the focus of Chapters 3 and 5 in this thesis. Under the guidance of the PhD supervisors, the PhD candidate drafted and re-worked the thesis chapters.
3 January 2018 Dr Carla Walton Date
ii Acknowledgements
The Acknowledgements are always my favourite part of any document and the part I read first. As a reader, they have given me some sense of the writer. As the writer, they are my chance to express gratitude to those who have supported me to complete this work. Some people have supported me directly with the research and the write- up, whilst others have supported me personally with various life challenges and in so doing, allowed me to keep on working on the thesis when my emotional energy was pulled elsewhere. Some people fell in both camps. I am extremely blessed. Over the years, I have written parts of the Acknowledgements in my head many times and it is thrilling to finally be writing them on the page.
To my primary supervisor, Amanda Baker, you are a formidable role model as a woman who is a Clinical Psychologist and who has trail blazed such a productive career as a clinical researcher. Thank you for your encouragement and support of me as an emerging clinical researcher in applying for grants and fellowships, whilst also helping me stay focussed on the importance of completing my PhD before I got too carried away with other things. I am very thankful for how responsive you have been with your feedback on my writing – I know how fortunate I am to have a supervisor who responds promptly and with supportive and thoughtful feedback. No matter where I was at, you would always finish our meetings or your emails with ‘well done’. You probably didn’t think much of it, but that praise was really encouraging and helped me to keep putting one foot in front of the other.
To my co-supervisor, Greg Carter, in the first instance, thank you for inviting me to DBT SPM and giving me the opportunity to enter into that world of colleagues
iii where I have learnt so much. Thank you for your belief in me that you have demonstrated through the many opportunities you have provided me. I am grateful for the thoroughness and the breadth of your feedback of my work. It challenged me to think more broadly and deeply about what I was communicating.
To my surrogate supervisor, Terry Lewin, this research would neither have commenced nor been completed if it weren't for you. From the outset, when others tried to dissuade us, you were encouraging of us embarking on the RCT and shared your wisdom with us in setting up the design of the study. Throughout the entire project, you were available and generous with your time in responding to phone calls / queries with ethical questions and day-to-day management issues. When it came to the analyses, I am extremely grateful for your wisdom, your calmness and your generosity. You stayed after the sun had gone down on many occasions with me, crunching things out in SPSS. And when I emailed you from various time zones where I was presenting the research at conferences and would send you a frantic email with some query in understanding the analyses (and when the sun had well and truly gone down in Australia), I would always wake in the morning to a calm, comprehensive email from you. And now even when you have been on long service leave and shifting towards retirement that you have still been willing and available to help me in these final stages has been awesome to me (in the true meaning of the word). Beyond the help related to this research project, you have been encouraging and supportive in your mentorship of me towards being a clinical researcher. I will never be able to thank you enough for your wisdom, your generosity and your kindness.
Nick Bendit, as my co-chief investigator on the RCT, we were in this together from the very beginning and all the way. Whilst we infuriated each other many times
iv in the process, I honestly don’t believe that I could have done it without you (although
I know that you say I could have). You made me get moving on things and let go of some of my perfectionistic tendencies which had the potential to ground the whole thing to a halt before it even got started. To carry out a large scale RCT (well large scale for BPD) in a routine clinical setting without a grant (initially) was an epic undertaking that brought with it many challenges. As those challenges arose, you were available to help me emotionally regulate to do the next necessary thing, and to pitch in whenever
I asked with some of the many hours of hard labour. I am extremely grateful to you for helping me bring to fruition this project that was meaningful, important and interesting to both of us.
To my mentor and friend, Kate Comtois, thank you for sharing your wisdom with me. I value your frankness and your generosity so much. I have grown as a clinician and as a researcher from the consultation sessions we have had. I love that I can geek out with you about research possibilities. When I have innovative ideas, I love that you will get enthused with me and start brainstorming and when they aren’t so innovative, that you will let me down gently.
Leslie Pollock, as Director of the service at the time the research was conducted, thank you for supporting the research and for your interest in it. Your support of the study on many levels, allowed this important research to happen.
I have been lucky enough to have had a number of very competent Research
Assistants who have been involved with the study. Amy Oswald, Maria O’Callaghan,
Stephanie Skinner and Victoria Wallis: your diligence with getting participants in for their assessments and the care with which you undertook the assessments and handled the data was critical to the overall quality of the study. I always knew that the
v research participants who gave so generously of their time and energy were in good hands with each of you. I have also been very grateful for how flexible you have each been in switching tasks with little notice as the moment required. I am also indebted to
Julia Dray, Julia Koller-Smith, Kate Bergin, Katie Dobinson, Kelly McDonnell, Katrina
Quick, Lisa Riley, Rebecca Stuart and Renee Shepherd, who volunteered their time to assist with data entry.
To the therapists who were involved in the study, all of whom I will list by name
(in alphabetical order): Agatha Conrad, Bridgette Lupton, Christopher Willcox, Craig
Hamilton, Dyani Nevile, Harsimrat Sandhu-Singh, Jane Middleby-Clements, Jennifer
Koorey, Jodie Fleming, Joy Herron, Jude Robinson, Karen Calabria, Kath McPhillips,
Katrina Bell, Kumudu Rhatnayaka, Leonie Funk, Leslie Pollock, Linda Bragg, Linda Kerr,
Lisa Blackwell, Maria Walker, Marianne Ayre, Megha Mulchandani, Natalie McCall,
Nick Bendit, Phoebe Webber, Stella Dyer, Subhra Bhattacharryya, Thomas Bellamy,
Yolandie Goodyear and Zoe Walker (and myself). I know that all of you care deeply about the people we serve and that those individuals were always your focus. The research at times probably felt like a pesky additional thing where you were given more forms to fill out when you were already busy enough. Thank you for making it possible for us to accumulate more knowledge that will help us to work towards improving what we offer to those who come to us. Chris, Leslie, Stella and Jennifer – thank you for sharing your input regarding the interface between the research study and the clinical setting.
To Tanya and Ayu, thank you for providing a space and the support to help unlock the writer in me at the Creative Connections writing retreat. In that first week away with you both, I overcame my internal blocks to writing and became a daily
vi writer from that point on. I am fortunate to have been paired with Pat Camp at that retreat who provided wise advice as well as good humour. Pat, thank you for offering the opportunity to get away from everything and to write at your holiday cabin one year after we first met. And, when I was hit by a car less than 24 hour after arriving in
Vancouver, thanks for the many ways you supported me to deal with all of the stuff that came with that in order to be able to get back to writing as soon as possible!
Professional editor Cameron Duder provided copyediting and proofreading services, according to the guidelines laid out in the university-endorsed national
‘Guidelines for editing research theses’. Cam, with your Mudge link, Kiwi background and excellent reputation, you were the perfect person to do the professional editing – thank you!
On a more personal note, to Mum and Dad, I am appreciative of the many sacrifices that you both made throughout my life in order for me to have the opportunity of a good education and to follow my heart’s desires. I have written much of this PhD on the desk that was bought with the money that Uncle Ted left for Kent and me to buy something for educational purposes. I think he would be pleased to know how much use it has had! Thank you for your generous gifts of service, which have freed me up at times to focus on the research without having to worry about other day-to-day things. And Mum, as for your suggestion that once I finish this PhD I can study medicine…um, well, no, I won’t be taking up that suggestion – this is enough formal study (at least for now!).
Most people I know who have completed PhDs have had various stressful events to contend with during the course of their candidature. I too had my share of emotionally challenging situations to contend with. Whilst I am fairly resilient, it is
vii impossible for me to imagine how I would have gotten through those challenges without an incredible group of human beings who supported me. To dear friends who helped in many aspects of my life, with their emotional support, with accountability, with encouragement and with reading drafts of chapters and giving comments. Thank you from the bottom of my heart, Anjie, Belinda, Ben, Brendon, Cath, Claire, Kathy,
Madonna, Michele, Michelle, Ness, Nick, Phoebe, Rachel and Shell. I want to single out
(double out perhaps?), Brendon and Phoebe, you have been an incredible cheer squad to have had on my side throughout this. Katie, thank you for stepping in and helping me in the final stage when my Dad died unexpectedly. I am very grateful for your practical help and your gentle, supportive manner.
In particular to Michele… it is hard to know how to thank you. I know you know how much you mean to me and what a massive difference your care and support has made in my life. I’m sure (fairly sure anyway), that I would have got to this point without your help, but it would have been a whole lot more of a struggle and with a whole lot less joy.
To Rose Elizabeth Walton, my niece, who is five-months-old at the time I am submitting this. You gave me motivation to finish this thing off and get it submitted once and for all so that I could spend hours face to face adoring you and doing everything possible to be the best Aunty I can be.
Finally, to the 162 participants of this study. Thank you for giving of your time and your emotional energy. Your participation allowed us to conduct this study and to contribute to the knowledge base of what is known about treatment of persons with
BPD in real world settings. This research was for you and for those that come after you
AND it could never have been conducted without you. We need to do so much more to
viii help reduce how miserable it is to live inside the skin of someone with BPD and to help you build lives that feel well and truly worth living. I was thrilled to see in our results that our interventions helped most of you in a range of areas. And yet, a portion of you discontinued treatment and a portion of you stayed the same or got worse with treatment. I commit to continue carrying out research that serves you.
ix Dedication
I dedicate this thesis to you, Mum and Dad.
Thank you for your love and support. I appreciate the sacrifices you both made to support my education. I wish you were both here for the completion of this degree. I
know you would have wanted to be here for it too (among many other things).
x List of grants and conference presentations relevant to this thesis
Grants
2017 University of Newcastle Health Behaviours Priority Research Centre
(PRC) travel grant. $2, 900
2017 University of Newcastle Brain and Mental Health PRC travel grant.
$1,000
2017 University of Newcastle Brain and Mental Health PRC statistical
support grant. $2,000
2016 Hunter New England Local Health District Health Research and
Translation Centre ‘RICH workshops grant’ for project entitled
‘Condensed Dialectical Behaviour Therapy (DBT) BPD: Pilot Study
exploring treatment outcomes for a reduced dose of treatment’.
$10,000.
2016 University of Newcastle Health Behaviours PRC for adherence coding
in Condensed DBT pilot study. $5,000
2016 University of Newcastle Brain and Mental Health PRC to fund Research
Assistant for ongoing follow-up data collection. $9,000
2011 Awarded Hunter New England Mental Health Service, Centre for Brain
and Mental Health ‘Research, Translation, Implementation and Models
of Care Program Grant’ for adherence coding in Randomised Clinical
Trial of Dialectical Behaviour Therapy and Conversational Model.
$10,000.
xi Oral Presentations
Walton, C.J., Bendit, N., Baker, A., Carter, G.L., & Lewin, T (2016, September).
Dialectical Behaviour Therapy and the Conversational Model in the treatment of
Borderline Personality Disorder: A randomised clinical trial in a public sector
mental health service. European Society for the Study of Personality Disorders
International Congress of Borderline Personality Disorder, Vienna, Austria.
Walton, C.J., Bendit, N., Baker, A., & Carter, G.L. & Lewin, T. (2016, June). Dialectical
Behaviour Therapy and the Conversational Model in the treatment of Borderline
Personality Disorder: A randomised clinical trial in a public sector mental health
service. Paper presented at the World Congress of Behavior and Cognitive
Therapies, Melbourne, Australia.
Walton, C.J., Bendit, N., Baker, A., & Carter, G.L. & Lewin, T. (2015, October). DBT and
the Conversational Model in the treatment of BPD: An RCT in a public sector
mental health service in Australia. Paper presented at the Dialectical Behavior
Therapy Strategic Planning Meeting, Seattle, USA.
Walton, C.J. (2015, September). Impact of Training & Clinician Allegiance on Adherence
to DBT in a public sector Mental Health team in Australia. Paper presented at
Seattle Implementation Research Conference, Seattle, USA.
Walton, C. J. (2013, July). Exploring factors that affect outcomes in clinical trials:
Discussion of a real world trial comparing Conversational and DBT models.
Invited speaker at Treatment of Borderline Personality Disorder: Promoting
Integration Conference, Wollongong, Australia.
xii Table of Contents
Acknowledgements ...... iii
Dedication ...... x
Table of Contents ...... xiii
List of Tables ...... xviii
List of Figures ...... xx
List of Abbreviations ...... xxi
Abstract ...... xxv
Chapter 1 Introduction ...... 1
1.1 Borderline Personality Disorder as a Clinical Disorder ...... 1
1.2 Treatments for BPD ...... 5
1.3 History and Overview of DBT and CM ...... 7
1.4 Efficacy vs. Effectiveness Research ...... 9
1.5 Efficacy Studies for Treatment of BPD ...... 11
1.5.1 Cognitive Behavioural Treatment of BPD ...... 47
1.5.2 Psychodynamic Treatment of BPD ...... 51
1.6 Effectiveness Studies for DBT and CM ...... 54
1.6.1 DBT Effecti veness Research ...... 81
1.6.2 CM Effectiveness Research ...... 85
1.7 Rationale for the Thesis ...... 88
1.7.1 Gaps in the Literature ...... 88
1.7.2 How the Current Study Will Address the Gaps ...... 94
1.8 Thesis Aims ...... 95
1.9 Overview of the Thesis...... 96
xiii Chapter 2 Methodology of a Randomised Clinical Trial of Dialectical Behaviour Therapy and Conversational Model ...... 98
2.1 The Current Study ...... 101
2.2 Methods/Design ...... 102
2.2.1 Study Design and Aims ...... 102
2.2.2 Study Setting ...... 103
2.2.3 Participants ...... 104
2.2.4 Content of the Interventions ...... 105
2.2.5 Therapists ...... 108
2.2.6 Training ...... 108
2.2.7 Procedure ...... 109
2.2.8 Withdrawal ...... 112
2.2.9 Assessments ...... 112
2.2.10 Treatment Adherence ...... 119
2.2.11 Sample Size and Data Analysis ...... 120
2.3 Discussion ...... 121
2.3.1 Strengths and Limitations ...... 122
Chapter 3 Newcastle Adherence Scale for Conversational Model (NASCOM): Development and Utility of an Adherence Measure for Conversational Model Psychotherapy ...... 125
3.1 Method ...... 135
3.1.1 Participants ...... 135
3.1.2 NASCOM Scale Development ...... 137
3.1.3 Procedure ...... 141
3.1.4 Statistical Analyses ...... 141
3.2 Results ...... 143
xiv 3.2.1 Inter-rater Agreement ...... 143
3.2.2 Discriminant Analysis ...... 147
3.2.3 NASCOM Adherence Cut-Off ...... 148
3.3 Discussion ...... 149
3.4 Conclusion ...... 153
Chapter 4 Dialectical Behaviour Therapy versus Conversational Model for Borderline Personality Disorder: Randomised Clinical Trial in a Public Sector Mental Health Service ...... 155
4.1 Method ...... 158
4.1.1 Study Design ...... 158
4.1.2 Participants ...... 159
4.1.3 Procedure ...... 160
4.1.4 Assessments ...... 163
4.1.5 Treatment ...... 165
4.1.6 Therapists ...... 167
4.1.7 Statistical Analysis...... 170
4.2 Results ...... 171
4.2.1 Recruitment ...... 171
4.2.2 Patient Characteristics ...... 173
4.2.3 Treatment Outcomes ...... 174
4.2.4 Comparison of Treatment Completers and Those Who Dropped Out .... 179
4.2.5 Fidelity ...... 179
4.3 Discussion ...... 180
4.3.1 Strengths and Limitations ...... 183
4.3.2 Future Research Implications ...... 186
xv Chapter 5 Comparing the Therapeutic Alliance across Dialectical Behaviour Therapy and the Conversational Model in the Treatment of Borderline Personality Disorder ...... 188
5.1 Method ...... 193
5.1.1 Participants ...... 193
5.1.2 Procedure ...... 194
5.1.3 Therapists and Research Assistants ...... 195
5.1.4 Treatments ...... 195
5.1.5 Measures ...... 196
5.1.6 Data Analysis ...... 197
5.2 Results ...... 199
5.2.1 Client-Rated Alliance (WAI) ...... 202
5.2.2 Therapist-rated Alliance (WAI) ...... 205
5.2.3 Client-rated Alliance and BPD Outcomes ...... 209
5.2.4 Therapist-rated Alliance and BPD Outcomes ...... 211
5.3 Discussion ...... 213
Chapter 6 Characterising the Clinical Impact of Dialectical Behaviour Therapy and the Conversational Model ...... 220
6.1 Method ...... 228
6.1.1 Participants ...... 228
6.1.2 Procedure ...... 229
6.1.3 Measures ...... 229
6.1.4 Data Analysis ...... 230
6.2 Results ...... 231
6.2.1 BPD Criteria ...... 231
6.2.2 Minimal Clinically Important Differences ...... 232
xvi 6.2.3 Mean number of outcomes of improvement and deterioration ...... 238
6.3 Discussion ...... 239
Chapter 7 Overall Summary and Future Directions ...... 244
7.1 Key Findings of the Thesis ...... 246
7.2 Integration of Key Findings of the Thesis ...... 250
7.3 Strengths and Limitations ...... 251
7.3.1 Strengths ...... 251
7.3.2 Limitations ...... 253
7.4 Future Directions ...... 255
7.5 Conclusion ...... 259
References ...... 260
Appendix A: Study Information Sheet ...... 290
Appendix B: Study Consent Form ...... 294
Appendix 3: Rater’s Manual for the Newcastle Adherence Scale for the Conversational Model (NASCoM) ...... 295
xvii List of Tables
Table 1.1 Participant Characteristics of Efficacy RCTs of Treatment of Borderline Personality Disorder (BPD) in Adults ...... 14 Table 1.2 Intervention Descriptions of Efficacy RCTs of Treatment of Borderline Personality Disorder (BPD) in Adults ...... 24 Table 1.3 Summary of Results of Efficacy RCTs of Treatment of Borderline Personality Disorder (BPD) in Adults ...... 34 Table 1.4 Participant Characteristics of Effectiveness Studies of Dialectical Behaviour Therapy (DBT) and Conversational Model (CM) for Treatment of Borderline Personality Disorder (BPD) in Adults ...... 56 Table 1.5 Intervention Descriptions of Effectiveness Studies of Dialectical Behaviour Therapy (DBT) and Conversational Model (CM) for Treatment of Borderline Personality Disorder (BPD) in Adults ...... 63 Table 1.6 Summary of Results of Effectiveness Studies of Dialectical Behaviour Therapy (DBT) and Conversational Model (CM) for Treatment of Borderline Personality Disorder (BPD) in Adults ...... 70 Table 1.7 Assessment of Risk of Bias of Studies ...... 92 Table 2.1 Summary of Measures...... 114 Table 3.1 Newcastle Adherence Scale for the Conversational Model (NASCOM) ...... 131 Table 3.2 Breakdown of Sessions Coded According to Therapy Model, Phase, and Therapists 140 Table 3.3 Psychometric Properties of Items in the NASCOM ...... 144 Table 3.4 Subscale and Total Score Profiles on the Two Adherence Scales ...... 148 Table 4.1 Baseline Sociodemographic and Clinical Characteristics of Participants ...... 173 Table 4.2 Generalised Linear Model Results of Co-Primary Outcomes of Suicide Attempts and NSSI and Depression Severity ...... 175 Table 4.3 Generalised Linear Model Results of Secondary Outcomes ...... 178 Table 5.1 Descriptive Statistics for Client and Therapist Alliance Rating (WAI) and BPD Outcomes by Assessment Period ...... 200 Table 5.2 Linear Mixed Model Results for Client and Therapist Alliance (WAI)—Significance of Effects ...... 202 Table 5.3 Linear Mixed Model Results for Client-Rated Alliance (WAI-C) and BPD Treatment Outcome—Significance of Effects ...... 210
xviii Table 5.4 Linear Mixed Model Results for Therapist-Rated Alliance (WAI-T) and BPD Treatment Outcomes—Significance of Effects ...... 212 Table 5.5 Client and Therapist Alliance Correlations ...... 213 Table 6.1 Characteristics of Individual Client-Level Outcomes of Minimally Clinically Important Differences (MCID) Analysis for Suicide Attempts and Non-Suicidal Self-Injury (NSSI) 234 Table 6.2 Characteristics of Client-Level Outcomes of Minimally Clinically Important Differences (MCID) Analysis for Continuous Variables ...... 236
xix List of Figures
Figure 2.1. Flow of Participants in the RCT ...... 111 Figure 3.1. Individual Session Total Scores on NASCOM and DBT Adherence Scales...... 149 Figure 4.1. Consort Diagram ...... 162
Figure 4.2. Mean No. of Suicide Attempts & NSSI Across T0-T2 Time Points by Condition ...... 176
Figure 4.3. Mean No. of Suicide Attempts Across T0-T2 Time Points by Condition ...... 176
Figure 4.4. Mean Number of Episodes of NSSI Across T0-T2 Time Points by Condition ...... 176
Figure 4.5. BDI-II Severity Scores Across T0-T2 Time Points by Condition ...... 177 Figure 5.1. Mean Client-Rated Total Alliance (WAI-C) Over Time ...... 202 Figure 5.2. Mean Client-Rated Tasks Over Time ...... 203 Figure 5.3. Mean Client-Rated Goals Over Time ...... 203 Figure 5.4. Mean Client-Rated Bond Over Time ...... 203 Figure 5.5. Client-Rated Total Alliance (WAI-C) Over Time...... 205 Figure 5.6. Client-Rated Alliance Components (Tasks, Goals, and Bond) Over Time ...... 205 Figure 5.7. Mean Therapist-Rated Total Alliance (WAI-T) Over Time ...... 207 Figure 5.8. Mean Therapist-Rated Tasks Over Time ...... 207 Figure 5.9. Mean Therapist-Rated Goals Over Time ...... 207 Figure 5.10. Mean Therapist-Rated Bond Over Time ...... 208 Figure 5.11. Therapist-Rated Total Alliance (WAI-T) Over Time ...... 209 Figure 5.12. Therapist-Rated Alliance Components (Tasks, Goals, and Bond) Over Time ...... 209 Figure 6.1. Profiles of Improvement Between Participants in Different Treatment Conditions Across All Domains ...... 238
xx List of Abbreviations
AAQ Acceptance and Action Questionnaire ADSHI Acts of Deliberate Self-Harm Inventory AES Spielberger Anger Expression Scale AIAQ Anger, Irritability and Assault Questionnaire BAI Beck Anxiety Inventory BASIS-32 Behaviour and Symptom Identification Scale BDI Beck Depression Inventory 1st edition BDI-II Beck Depression Inventory 2nd edition BDQ Brief Disability Questionnaire BEST Borderline Evaluation of Severity over Time BHS Beck Scale for Hopelessness BIS Barratt Impulsiveness Scale BPD Borderline Personality Disorder BPD-40 Borderline Personality Disorder Checklist-40 BPD-TOA Brief Borderline Personality Disorder Treatment Outcome Assessment BPDSI-IV Borderline Personality Disorder Severity Index 4th edition BPRS Brief Psychiatric Rating Scale BPSDI Borderline Personality Disorder Symptom Index BSI Borderline Symptom Inventory BSL Borderline Symptom List BSSI Beck Scale for Suicidal Ideation CBT Cognitive Behaviour Therapy CGI Clinical Global Impression CISSB Cornell Interview for Suicidal and Self-Harming Behaviour CM Conversational Model COM Treatment Completers CORE-OM Clinical Outcomes in Routine Evaluation – outcome measurement CP Community Psychotherapy
xxi CT Cognitive Therapy DASS Depression Anxiety Stress Scale DASS-D Depression Anxiety Stress Scale – Depression Subscale DBT Dialectical Behaviour Therapy DERS Difficulties in Emotion Regulation Scale DES Dissociative Experiences Scale DIB-R Diagnostic Interview for Borderline Personality Disorders – Revised DPRS Derogatis Psychiatric Rating Scale DSHI Deliberate Self-Harm Inventory DSS Dissociation Tension Scale EQ-5D EUROQOL Quality of Life Measure ERGT Emotion Regulation Group Therapy GAF Global Assessment of Functioning GAF-F Global Assessment of Functioning - Social Function Subscale GAF-S Global Assessment of Functioning – Symptoms Subscale GPM General Psychiatric Management / Good Psychiatric Management GSA Global Social Adjustment GSI Global Severity Index HAM-D Hamilton-Depression Scale HARS Hamilton Anxiety Rating Scale IIP Inventory of Interpersonal Problems IVE Eysenck Impulsivity Venturesomeness Empathy K10+ Kessler 10 Scale KABOSS-S Karolinska Affective and Borderline Symptom Scale – Self- Assessment KIMS Kentucky Inventory of Mindfulness Skills LPC Lifetime Parasuicide Count MANSA Manchester Short Assessment of Quality of Life MBT Mentalisation Based Therapy MCMI-III Millon Clinical Multiaxial Inventory, Third Edition
xxii MSI-BPD McLean Screening Instrument for Borderline Personality Disorder NSSI Non-Suicidal Self-Injury OAS Overt Aggression Scale-Modified PHI Parasuicide History Interview PSDI Positive Symptom Distress Index PST Positive Symptom Total QOLI Quality of Life Inventory QTF Questionnaire of Thoughts and Feelings RCT Randomised Clinical Trial / Randomised Controlled Trial RLI Reasons for Living Inventory RST Rogerian Supportive Therapy SAS Social Adjustment Scale SASI-Count Suicide Attempt and Self-Injury Count SASII Suicide Attempt Self Injury Interview SCID-BPD Structured Clinical Interview for DSM-IV, Personality Disorders, Borderline Personality Disorder criteria SCL-90-R Symptom Checklist 90 Revised SCM Structured Clinical Management SFQ Social Functioning Questionnaire SFT Schema Focused Therapy SGT Supportive Group Treatment SSHI Suicide and Self-Harm Inventory SSI Sense of Self-Injury STAI-S State Trait Anxiety Inventory – State Anxiety STAI-T State Trait Anxiety Inventory – Trait Anxiety STAXI Speilberger Anger Scale STEPPS Systems Training for Emotional Predictability and Problem Solving STIPO Structured Interview of Personality Organization
T0 Baseline Assessment point
T1 Mid-treatment assessment point (7 months)
xxiii T2 Post-treatment assessment point (14 months) TAU Treatment as Usual TFP Transference Focused Psychotherapy THI Treatment History Interview WAI Working Alliance Inventory WHOQOL World Health Organization Quality of Life Assessment WHOQOL-BREF World Health Organization Quality of Life Assessment – Brief form WL Waitlist YSQ Young Schema Questionnaire ZRSB Zanarini Rating Scale for Borderline Personality Disorder
xxiv Abstract
Borderline Personality Disorder (BPD) is a disabling mental disorder that is associated with a high degree of suffering for the individual. Large-scale studies have shown pervasive social and functional impairment. It is associated with intensive use of mental health services and is recognised as a challenging disorder for clinicians to treat. There was previously little hope about the capacity for BPD to be successfully treated. In the past 20 years, there has been considerable progress in psychotherapeutic treatments developed and evaluated for BPD. Psychotherapy, rather than psychiatric medication is the indicated treatment for BPD. There are a number of psychotherapies that have been developed specifically for the treatment of adults with a diagnosis of BPD, with Dialectical Behaviour Therapy (DBT) being the therapy with the greatest evidence base. However, the outcome research for BPD is in its infancy. There are a number of limitations in the existing randomised studies. Many have small sample sizes; apart from DBT, most treatment models have only been evaluated in one or two studies and the majority of studies have been conducted by treatment developers or investigators who are strongly allegiant to one particular model of therapy. DBT has been evaluated in a number of efficacy studies but few effectiveness studies. It has rarely been compared against another active treatment for
BPD. Other therapies for BPD, such as the Conversational Model (CM), a psychodynamic therapy, show promising results. However, CM has not been investigated in a randomised trial, nor has it been evaluated against another evidence- based treatment for BPD.
This thesis describes the methodology and outcomes of a randomised clinical trial (RCT) conducted in a public sector mental health service comparing DBT and CM in
xxv the treatment of suicidal and non-suicidal self-injurious behaviour and depression severity among persons with BPD. The methodology of the trial is described in Chapter
2. Persons with a diagnosis of BPD and recent suicidal and non-suicidal self-injury were randomised to receive either DBT or CM for 14 months. Outcomes were evaluated at baseline, mid-treatment (7 months) and post-treatment (14 months). Chapter 3 discusses the development and evaluation of an adherence measure for CM. The tool was found to have good inter-rater reliability on items and to clearly discriminate between the two treatments. Outcomes from the RCT are discussed in Chapter 4. Both treatments showed significant improvement over time across the 14 months of therapy in suicidal and non-suicidal self-injury and depression scores. There were no significant differences between the treatment models in reduction of suicidal and non- suicidal self-injury. However, DBT was associated with significantly greater reductions in depression scores compared to CM.
The pattern of results was similar with the secondary outcomes such that scores on BPD severity, dissociation, interpersonal problems, sense of self, mindfulness capacity and difficulties in emotion regulation all significantly improved with both treatments. At the differential level, DBT was associated with significantly better improvement in mindfulness capacity and emotion regulation. Chapter 5 reports on the findings of the RCT, in relation to the working alliance. The therapeutic alliance in
DBT and CM was compared for therapist-rated and client-rated alliance overall, as well as distinct components of the alliance in the sub-scales of task, goal, and bond. There was a significant treatment effect overall for client-rated goals, which were significantly greater in DBT than CM. Therapist-rated goals were also significantly greater in DBT than CM in a time by treatment interaction effect. Chapter 6 reports on
xxvi changes at the individual level, beyond the aggregated results by treatment group reported in Chapters 4 and 5. The majority of participants improved in terms of their suicidal and non-suicidal self-injury, severity of BPD symptoms, depression scores, and dissociation scores. Despite this improvement in severity of symptoms, only 38% of the sample no longer met criteria for BPD after 14 months of treatment. The majority of participants fell in the ‘unchanged’ category on interpersonal problems, sense of self and mindfulness capacity. The final chapter of this thesis synthesises the main findings of the preceding six chapters. This research adds to the accumulating body of knowledge of psychotherapeutic treatment of BPD and supports the use of both DBT and CM as effective treatments in routine clinical settings, with some additional benefits for DBT for persons with co-morbid depression. Future research directions are identified and discussed.
xxvii
Chapter 1 Introduction
This thesis describes a randomised clinical trial (RCT) and associated components of that trial, conducted in a specialist service for Borderline Personality
Disorder (BPD) within public mental health in Newcastle, Australia. This opening chapter will include information regarding BPD as a clinical disorder, the psychotherapeutic interventions that have been developed so far to treat persons with
BPD, and the evidence base for these interventions. There is particular emphasis on the two psychotherapies that are the focus of this research, Dialectical Behaviour
Therapy (DBT) and the Conversational Model (CM). Given the real world setting of the current research, this chapter will include detailed information about the studies that have been conducted in routine clinical settings as a well as a summary of risk of bias for each study. There have not been any randomised trials evaluating CM for BPD and there have been very few RCTs using active treatment comparison groups for DBT, particularly in routine clinical settings. The current study will address those gaps. This chapter will conclude with the overall aims of the research and an overview of the remaining chapters in the thesis.
1.1 Borderline Personality Disorder as a Clinical Disorder
BPD is a disabling mental disorder characterised by instability in a range of domains including affect, interpersonal relationships, identity, and behavioural dysregulation (Lieb, Zanarini, Schmahl, Linehan, & Bohus, 2004). It is associated with a
1
high degree of suffering, efforts to deliberately inflict harm to oneself, high rates of suicide attempts, and a lifetime suicide mortality rate of approximately 10% (Black,
Blum, Pfohl, & Hale, 2004; Pompili, Girardi, Ruberto, & Tatarelli, 2005). Prevalence of lifetime BPD in the USA has been identified as 5.9% (Grant et al., 2008). At least 50% of chronically suicidal patients with four or more visits in a year to a psychiatric emergency department are persons with a diagnosis of BPD (Bongar, Peterson,
Golann, & Hardiman, 1990) and it is the most commonly diagnosed personality disorder in both inpatient and outpatient settings (Trull, Freeman, Vebares, Choate,
Helle, & Wycoff, 2018). In order to meet diagnostic criteria for BPD, a person needs to meet five out of the following nine criteria: (1) frantic efforts to avoid real or imagined abandonment; (2) a pattern of unstable and intense interpersonal relationships characterised by alternating between extremes of idealisation and devaluation; (3) identity disturbance: markedly and persistently unstable self-image or sense of self; (4) impulsivity in at least two areas that are potentially self-damaging (e.g., spending, sex, substance abuse, reckless driving, binge eating); (5) recurrent suicidal behaviour, gestures or threats or self-mutilating behaviour; (6) affective instability due to a marked reactivity of mood (e.g., intense episodic dysphoria, irritability, or anxiety usually lasting a few hours and only rarely more than a few days); (7) chronic feelings of emptiness; (8) inappropriate, intense anger or difficulty controlling anger (e.g., frequent displays of temper, constant anger, recurrent physical fights); and (9) transient, stress-related paranoid ideation or severe dissociative symptoms (American
Psychiatric Association, 1994).
Criterion 5 refers to recurrent suicidal behaviour, gestures or threats or self- mutilating behaviour. This behaviour may be with the intention to die, without the
2
intention to die, or ambivalence (Linehan, Comtois, Brown, Heard, & Wagner, 2006a).
It is not always clear what the function of the behaviour of harming oneself is and hence, which of these classifications the behaviour falls into. Within the scientific community, there is ongoing debate about how to correctly define self-harming behaviours (Kapur, Cooper, O’Connor, & Hawton, 2013) with a range of terms used, including self-harm, non-suicidal self-injury, self-directed violence, or self-injury.
Broadly, non-suicidal self-injury (NSSI) refers to directly harmful behaviours without suicidal intent (Plener, Schumacher, Munz, & Groschwitz, 2015). However, NSSI does not usually include deliberate self-poisoning (Kapur et al., 2013), despite self-poisoning occurring at times without suicidal intent, for example, a situation where an individual takes more than a prescribed dose of a medication in order to escape their emotions.
In the current thesis, NSSI was the term chosen to refer to self-harm without the intention to die that encompasses a range of methods, such as cutting or burning, as well as deliberate self-poisoning. This is in keeping with the clinician-administered measure used to assess self-harm in the current study, the Suicide Attempt Self-Injury
Count (SASI-Count; Linehan & Comtois, 1996), and other seminal BPD outcomes studies (Linehan et al., 2006b; McMain et al., 2009).
Trull et al. (2018) note the significant limitations with the categorical system of diagnosis of BPD, including marked variability within categories, arbitrary cut-off points and high levels of co-morbidity with Axis I and Axis II disorders. The rate of comorbidity with Axis I mental disorders, has been estimated at 85% (Leichsenring, Leibing, Kruse,
New, & Leweke, 2011) making co-morbidity more the rule than the exception. In addition, most patients diagnosed with a personality disorder meet criteria for more than one personality disorder. These high rates of co-occurrence can be a barrier in
3
clinical practice in terms of clarity about treatment recommendations (Morey et al.,
2015). Given the limitations of categorizing BPD, there is a push from some experts for a dimensional approach whereby BPD represents a level of personality organisation or dysfunction that cuts across diagnostic categories (Trull, 2018).
Beyond high levels of symptomatic impairment, large-scale studies have shown pervasive social and functional impairment (Gunderson et al., 2011). The broad subjective experience of BPD patients is of feeling overwhelmed, worthless, angry, lonely, and misunderstood (Zanarini et al., 1998). It is considered to be the most common personality disorder in clinical populations and is associated with intensive use of mental health services. While community BPD prevalence rates are estimated to be around 3% (Tomko et al., 2013), they are higher within mental health settings, where approximately 10% of all psychiatric outpatients and 15–25% of psychiatric inpatients meet the criteria for BPD (Leichsenring et al., 2011).
The diagnosis of BPD was first entered in the American Psychiatric Association’s
DSM-III in 1980 (American Psychiatric Association, 1994) and since that time has grown markedly in recognition and use (Gunderson & Links, 2008). Many clinicians and researchers have described BPD as one of the most challenging psychiatric disorders to treat (Bateman & Fonagy, 2000; Linehan, 1993a) and one that evokes high stigma amongst mental health practitioners (Aviram, Brodsky, & Stanley, 2006). Until the
1990s, there was little hope about the capacity for BPD to be successfully treated
(Korner & McLean, 2017). In the past 20 years, there has been considerable progress in psychotherapeutic treatments developed and evaluated for BPD, and most have been shown to be more effective than control interventions for BPD-related problems, such as suicidality and NSSI (Cristea et al., 2017). Most patients with BPD will experience
4
remission of symptoms with treatment (Gunderson, 2015). Psychotherapy rather than psychiatric medication is the indicated treatment for BPD (National Health and Medical
Research Council, 2012).
1.2 Treatments for BPD
There are a number of psychotherapies that have been developed specifically for the treatment of adults with a diagnosis of BPD. These treatments can be considered under the umbrella of cognitive behavioural or psychodynamic therapies.
These treatments have shown positive results in the treatment of BPD, in particular with reductions in suicidal and NSSI behaviour, and hospital admissions (Choi-Kain,
Albert, & Gunderson, 2016; Stoffers et al., 2012).
Of the cognitive behavioural models, the therapy that is most widely recognised (Gunderson, 2015) and has the most empirical support for BPD is DBT
(Stoffers et al., 2012). Others include Systems Training for Emotional Predictability and
Problem Solving (Black et al., 2008), Schema Focused Therapy (Farrell, Shaw, &
Webber, 2009) and Cognitive Therapy (Davidson et al., 2006). The research evidence for each will be described separately in section 1.5.1, “Cognitive Behavioural
Treatment of BPD.”
There is also empirical evidence for a number of psychodynamic therapies, namely Mentalisation Based Therapy (Bateman & Fonagy, 1999, 2009); Transference
Focused Therapy (Clarkin, Levy, Lenzenweger, & Kernberg, 2007; Doering et al., 2010); the American Psychiatric Association’s Good Psychiatric Management (previously referred to as General Psychiatric Management; McMain et al., 2009); Dynamic
Deconstructive Psychotherapy (Gregory & Ramen, 2008); and CM (Korner, Gerull,
5
Meares, & Stevenson, 2006; Meares, Stevenson, & Comerford, 1999; Stevenson &
Meares, 1992; Stevenson, Meares, & D’Angelo, 2005). The research evidence for each will be described separately in section 1.5.2, “Psychodynamic Treatment of BPD.”
Research presented in this thesis will focus on two of these treatments: DBT, a type of CBT, and CM, a psychodynamic treatment. DBT and CM were the first treatments for BPD with published research about their outcomes (Linehan,
Armstrong, Suarez, Allmon, & Heard, 1991; Stevenson & Meares, 1992). The rationale for focusing on these two treatments of all of the evidence-based treatments developed for BPD will be briefly outlined. DBT is currently the treatment with the most empirical support (Cristea et al., 2017); it is used in many services across
Australia (Carter, Willcox, Lewin, Conrad, & Bendit, 2010; Pasieczny & Connor, 2011;
Prendergast & McCausland, 2007; Williams, Hartstone, & Denson, 2010) and it is easy to teach and learn (Swenson, 2000). CM for BPD, while having much less empirical support, was largely developed in Australia (Korner and McLean, 2017). The training program for CM is run by one of the few Australian and New Zealand training institutes that offer specialised training for BPD (Haliburn, Stevenson, & Gerull, 2009). Although the findings from the initial studies were promising, they employed non-randomised designs, and the findings have not been replicated. It is a treatment model used commonly in Australia, despite the lack of empirical support and on these grounds it seemed important to examine it using an RCT design. In order to provide the reader with some context, further information will be provided about these two forms of treatment prior to a more detailed discussion of the research evidence for the range of treatments for BPD.
6
1.3 History and Overview of DBT and CM
DBT proposes that the core psychopathology of borderline patients involves an interaction between a biological sensitivity and an emotionally invalidating environment that results in patients failing to develop a capacity to regulate their emotions (Linehan, 1993). This theoretical position evolved from Marsha Linehan’s work with suicidal patients who were responding poorly to a purely behavioural model of treatment. Standard DBT is a manualised treatment developed by Professor Linehan
(1993a) that combines treatment strategies from behavioural, cognitive, and supportive psychotherapies. It includes weekly individual sessions in a pre-treatment phase to orient participants to the treatment model, explore goals, and elicit a commitment to therapy. Pre-treatment sessions are followed by concurrent weekly individual therapy and skills training group sessions for 12 months. The individual therapy sessions take approximately one hour per week and apply directive, problem- oriented techniques (including behavioural skill training, contingency management, and cognitive modification) alongside supportive techniques. The philosophy of dialectics underpins the therapy (Linehan, 1993a). The skills training groups meet weekly for 2.5 hours and follow a psycho-educational format. Behavioural skills in three main areas are taught in three modules: interpersonal effectiveness, distress tolerance, and emotion regulation skills. Preceding each module is a two-week focus on mindfulness, which is the core skill taught over the entire therapy period. The skills group includes the teaching and application of skills and the review of home practice between sessions. In addition to attending individual therapy and skills training groups, participants have access to telephone coaching to generalise skills learned in face-to- face sessions to broader contexts. At the service where this research was conducted,
7
the telephone coaching function of the therapy occurs in working hours by accessing the patient’s individual therapist by telephone. Outside of therapist working hours, participants have access to a telephone service staffed by a roster of six DBT therapists, which provides phone coaching to generalise skills learned in therapy in participants’ lives. DBT therapists also attend a weekly consultation team meeting designed to provide support and assist therapists to be adherent to the treatment model.
CM was developed by Robert Hobson and Russell Meares and involves an integration of experiential/humanist, interpersonal, and psychodynamic interventions
(Shapiro & Startup, 1992). While Hobson’s and Meares’ early focus was on interpersonal disturbance and its impact on mental well-being, Meares (2004) later adapted CM to those persons with a diagnosis of BPD. Meares was interested in the internal experience of clients with BPD, and the damage to their sense of “self,” which he describes as an internal flow of feelings, thoughts, and meanings. From his point of view, a healthy self enables the person to experience intimacy in relationships and retain the capacity to experience aloneness. This sense of self is compromised or stunted in BPD, particularly at times of high emotional arousal. The task of the therapist in CM is to attune to the emotional experience of the client, so as to develop a shared understanding of the internal feelings, thoughts, and sensations of the client.
This enables the client to make sense of their behaviour in the world, particularly in close relationships, both past and present. The therapy centres on the therapeutic relationship as a template for other relationships and aims to help individuals increase their capacity for reflective functioning. The treatment model is outlined in a published treatment manual (Meares, 2012). The individual therapy sessions are approximately
8
of one hour and are nondirective. The focus is on understanding the patient’s emotional experience and actively describing that experience back to the patient. The key intervention for change in CM is the therapist staying in the “here and now” of therapy by focusing on client feelings (stated and unstated). The therapist actively looks for subtle signs of emotional misunderstanding in the patient, leading to mutual self-reflection and repair of the moment of disconnection in the therapeutic relationship. High value is placed on the patient’s real experience (versus socially acceptable experience) and the development of an authentic personal narrative. The patient is encouraged to find links between the maladaptive relationship patterns they have developed in their current social world and the relationship pattern they have with the therapist (and possibly, but not necessarily, the links with their childhood relationships). The name of the therapy refers to the idea of the above interventions occurring within a natural conversation between the therapist and the client.
1.4 Efficacy vs. Effectiveness Research
Prior to considering the scientific evidence for the psychotherapeutic treatments designed for BPD, the different types of research models will be outlined.
Evidence-based psychological treatments are usually developed and evaluated under ideal conditions in controlled settings. This allows researchers to establish internal validity of the therapy and to demonstrate that the changes that occur are due to the effects of the therapy and not to chance or other factors such as the passage of time
(Chambless & Hollon, 1998). Outcome research conducted under these tightly controlled conditions is known as efficacy research. Features of efficacy trials include the random assignment of patients to treatments, assessment of outcomes by
9
someone blind to the treatment condition, use of objective outcome measures, use of treatment manuals, and monitoring of treatments delivered to ensure fidelity (Carroll
& Rounsaville, 2003).
However, evidence-based therapies are developed and evaluated with the ultimate goal of broad use for persons presenting for help in routine clinical settings. In order to determine if the results of efficacy studies generalise to applied clinical settings (and how well they generalise), therapies are tested in these applied settings
(Borkovec & Castonguay, 1998). These trials are known as effectiveness trials. A number of components should be included to test whether treatments are feasible in routine clinical settings and whether they have measurable positive effects.
Effectiveness research typically includes greater diversity in patients and clinicians, such that clinical populations are included with comorbid psychopathology and varying durations of illness. Clinicians in these studies do not necessarily have advanced degrees, are generally not closely supervised by experts, nor are their sessions monitored for how closely they adhere to the treatment protocol (Nathan, Stuart, &
Dolan, 2000). Evaluating whether treatments are effective, without the tight restrictions inherent in efficacy trials, provides vital evidence about the real world benefit of these treatments. Effectiveness studies examine how treatments hold up when delivered to usual clinical populations by routine therapists in real world settings
(Chambless & Hollon, 1998).
A third type of research design has been proposed which retains critical features of efficacy trials while adding aspects from effectiveness design, and this has been referred to as a hybrid efficacy-effectiveness design (Atkins, Frazier, & Cappella,
2006; Carroll & Rounsaville, 2003). Many clinical trials fall somewhere on the
10
continuum between an efficacy and an effectiveness design, although they are rarely explicitly referred to as hybrid designs. The benefit of a hybrid design is that it preserves scientific rigour while expanding the study to include issues of interest to clinicians and policy makers with the aim of translating research into practice more promptly and more effectively. Further, a hybrid design helps bridge the gap between efficacy and effectiveness research (Glasgow, Lichtenstein, & Marcus, 2003). The current study utilised a hybrid efficacy-effectiveness design.
1.5 Efficacy Studies for Treatment of BPD
Studies reviewed in the following sections included adults with a diagnosis of
BPD where the treatment was delivered in an outpatient setting. In order to identify relevant studies two online databases were searched: Medline and PsycInfo using the search terms “Borderline Personality Disorder” AND “treatment” OR “psychotherapy”
OR “therapy” OR “psychological intervention” AND restricted to English language publications. No limits were placed on publication year.
Studies were also identified from a recent systematic review and meta-analysis
(Cristea et al., 2017) which included studies where an independent psychotherapy was compared to a control intervention or where an experimental condition was added to usual treatment. Studies comparing active psychotherapies were identified from the
Cochrane Database of Systematic Reviews “Psychological Therapies for People with
BPD” (Stoffers et al., 2012). Further studies were identified by searching the reference lists of the studies identified in the literature review and in the Cochrane database to identify any other published studies of treatment of adults with BPD in outpatient settings that had been overlooked. Results of studies are reported for RCTs of
11
psychological therapies of BPD in adults where there is at least one replication study.
Studies were identified as efficacy studies (as opposed to effectiveness studies covered in the next section) based on criteria outlined in Flay et al. (2005) and Gartlehner,
Hansen, Nissman, Lohr, and Carey (2006). Information about the study settings, inclusion and exclusion criteria and participant characteristics is included in Table 1.1.
Of note is that most studies are conducted with white females. Most studies excluded people with schizophrenia, bipolar disorder or a cognitive impairment or substance dependence (but not substance abuse). Descriptions of interventions evaluated in each study, including information about the control conditions, are shown in Table 1.2.
Information regarding attrition and the outcomes for each study are reported in Table
1.3. Due to marked heterogeneity in outcomes measured across studies, outcomes included were those that were commonly used in BPD outcome research. As seen in
Table 1.3, there were 20 efficacy RCTs; 12 of these studies compared an active treatment with a control condition, such as Treatment as Usual (TAU); 5 of these studies compared an active treatment with a more stringent control condition, such as community experts; and 3 of these studies compared an active treatment against another active treatment. Of the 12 studies where an active treatment was compared against TAU, eight reported at least 50% of outcomes significantly in favour of the intervention. Of the eight studies that compared an active treatment with either an active control or another active treatment, only one study showed significant differences for one of the interventions over the others on at least 50% of outcomes.
Follow-up data was collected in 11 of the 20 studies. Details of follow-up data are included in the table for completeness, however, all significance levels and effect
12
sizes are based on the post-treatment period, as identified by the authors of each of the studies.
13
Table 1.1 Participant Characteristics of Efficacy RCTs of Treatment of Borderline Personality Disorder (BPD) in Adults Authors N Mean % % Ethnicity Education Inclusion Exclusion criteria Illness factors Country in Female Welfare criteria years Marital Status (SD) Bateman 38 MBT = 58% Not reported. MBT: college education Aged 16 to 65 Met criteria for All participants were recruited from (1999) 30.3 36%; Unemployed 100%; years, met schizophrenia, the Halliwick Psychotherapy Unit, (5.86) Single 89%; sheltered criteria for bipolar disorder, referred during 1993 and 1994. U.K. accommodation 16% BPD in substance misuse, TAU = accordance mental Comorbid Axis I diagnoses: Major 33.3 Control patients: college with DSM-III- impairment, or depression 65.79%, dysthymia (6.60) education 16%; TR, score of >7 evidence of 57.89%, panic disorder 52.63%, 14 unemployed 100%; on the DIBR organic brain agoraphobia 36.84%, social phobia single 84%; sheltered disorder. 13.16%, bulimia 31.58% accommodation 10%
Bateman 134 MBT = 80% MBT: White MBT Group: tertiary Aged 18 to 65 Currently in long- All participants were recruited (2009) 31.3 British/Europea education 36.6%; years, meet term through referrals for personality (7.6) n 76.1%; Black Currently employed diagnostic psychotherapy, disorder treatment from clinical U.K. African/Afro- 28.2%; state benefit criteria for met criteria for services during 2003 to 2006. SCM = Caribbean 66.2%; Married 19.7% BPD, >suicide psychotic disorder 30.9 15.5%; Other attempt or or bipolar I Comorbid Axis I diagnoses: Major (7.9) 8.5% SCM Group: tertiary episode of disorder, opiate depressive disorder 55.97%, education 42.9%; life- dependence depressive disorders/dysthymia SCM: White Currently employed threatening requiring specialist 76.87%, PTSD 14.18%, any anxiety British/Europea 30.2%; state benefit self-harm treatment, mental disorder 61.19%, any substance use n 68.3%; Black 68.3%; Married 9.5% within 6 impairment or disorder 53.73%, any eating disorder African/Afro- months of evidence of 27.61%, somatoform disorder Caribbean participation organic brain 12.69%. Current axis II disorder: 20.6%; Other disorder Cluster A 41.79%, Cluster B 51.49%, 11.1% Cluster C 38.06%
Blum 124 31.5 83% Caucasian 94% Graduate degree 6%; DSM-IV Non-English Participants were recruited from (2008) (9.5) African college degree 16%; diagnosis of speakers, University of Iowa mental health American 2% some college 52%; high BPD psychotic or services, referred from local metal USA Other 3% school 20%; 15 Bos 79 STEPPS 86% Not reported. STEPS: >high school Adults (18+ Non-Dutch All participants recruited from non- (2010) = 32.9 7.2%; high school 8.1%; years) who speaking, cognitive academic outpatient clinics of two (5.6) middle school 38.1%; met criteria impairment mental health care institutes in the The elementary school for BPD in (IQ<70), treated Netherlands. Nether- TAU = 16.7%. accordance involuntarily, lands 31.8 Employed 19.0%; with DSM-IV presented Comorbid diagnoses not reported. (9.2) disabled 38.1%; Other criteria, imminent danger 42.9%. exceeding cut- to self or others Having a partner 52.4%. off scores on either or both TAU: >high school 2.7%; the impulsivity high school 29.7%; and middle school 51.4%; parasuicide elementary school subscales of 16.2%. the BPDSI-IV Employed 13.5%; disabled 43.2%; Other 43.2%. Having a partner 67.6%. Clarkin 90 30.9 92% Caucasian Graduate training Met DSM-IV Comorbid Participants with a reliable diagnosis (2007) (7.85) 67.8%; African 18.9%; college degree criteria for psychotic of BPD were recruited from New American 32.2%; associate’s diagnosis of disorders, bipolar I York City and the adjacent tri-state USA 10.0%; Hispanic degree 6.7%; some BPD, adults disorder, area. 8.9%; Asian college 31.1%; high between 18 delusional 5.6%; Other school graduate 7.8%; and 50 years disorder, delirium, Lifetime axis I disorder comorbidity 7.8% 16 (2009) 34.3 54.5%; single 48.5% criteria for age, psychotic two university hospital centers; Lyon (10.2) diagnosis of disorder(s) with Anxiety Disorder Unit, and Marseille RST: university diploma BPD, score of current delusions, Behaviour Therapy Unit RST = 53.1%; single 53.1% >8 on DIBR significant drug or Axis I disorder comorbidity statistics: 32.6 alcohol addiction, 40% social phobia, 26% panic (8.3) antisocial disorder, 15% agoraphobia, 18% behaviours, living PTSD (current), 15% hypomania, significantly far 28% bulimia nervosa, 56% from the generalized anxiety disorder, 55% treatment centers major depressive disorder (current). Davidson 108 31.9 84% White 100% Postgraduate degree 18 to 65 yrs of Current inpatient Participants were referred by (2006) (9.1) 2.8%; first degree 6.6%; age, met treatment for clinicians across a range of services 1-3 years college training criteria for > 5 mental state (e.g., community mental health 28.3%; 5-6 years items for BPD disorder, receiving teams, clinical psychology and secondary school 8.5%; on SCID-II for a systematic liaison psychiatry services) left school at 16 years DSM-IV Axis II psychotherapy or 38.7%; Other 15.1%; PDs, received specialist service Comorbidity statistics not reported. Special educational either (e.g. needs at school 17.9% inpatient psychodynamic psychiatric psychotherapy), Living alone 25.8%; services OR an insufficient English unemployed 67.9%; any assessment at language benefits 84%; involved emergency proficiency, living with any legal/law services OR an far from treatment enforcement services episode of facilities, organic during last 6mths 47.2% self-harm illness, mental within the impairment, Single 79.2%; married previous 12 alcohol/drug 13.2%; divorced 7.6% mths, ability dependence, to provide schizophrenia or informed bipolar affective consent disorder Doering 104 TFP = 100% Not reported. TFP (CP in brackets): Female, 18 to Comorbid ASPD, Participants recruited from (2010) 27.46 No compulsory school 45 years, schizophrenia, outpatient units from Psychiatry and 17 (6.8) 15.4%(15.4%); diagnosis of bipolar I and II Psychotherapy Departments of local Austria compulsory school BPD in disorder with a universities. CP = 9.6%(15.4%); accordance major depressive, 27.19 apprenticeship with DSM-IV manic or The average number of Axis I (7.5) 21.1%(32.7%); A-level criteria, hypomanic diagnoses among the TFP group was 50%(36.5%); Academic sufficient episode during the 1.6, and among the CP group 1.5, 9.6%(9.6%); still in German previous 6 months, The average number of Axis II school 5.8%(1.9%) language substance diagnoses in the TFP group was 2.5, In occupational training proficiency dependency among the CP group 2.2 34.6%(32.7%); during the unemployed previous 6 months, 19.2%(21.2%); part-time organic pathology 11.5%(13.5%); full-time or mental 26.9%(28.8%); retired retardation 5.8%(7.7%). Single 50% (44.2%); unmarried with partner 30.8%(34.6%); married 13.5%(13.5%); divorced 5.8%(7.7%) Farrell 32 SFT = 100% Not reported. College graduate Met criteria Axis I diagnosis of All participants were referred by (2009) 35.5 31.25%; some college for BPD in a psychotic their individual (9.30) 37.5%; high school accordance disorder, IQ <89 psychotherapist/psychologist USA graduate 18.75% with DIBR and providing outpatient community Control Housewife 12.5%; the Borderline services. = 35.9 student 9.38%; Syndrome (8.08) employed 53.13%, Index, adult Comorbidity statistics not reported. disability 9.38%, females aged supplemental security between 18 income 3.13% and 65 years Married 7.7%; divorced 44.4%; living with partner 12.2%; in relationship 23.3% 18 Giesen- 88 SFT = 93% Not reported. Graduate/professional Met DSM-IV Psychotic disorder Participants were referred from Bloo 31.70 11.63%; college graduate criteria for (excepting short, therapists at secondary and tertiary (2006) (8.9) 11.63%; some college primary reactive psychotic local community mental health 36.05%; high school diagnosis of episodes), bipolar institutes. The TFP = graduate 17.44%; grades BPD, adults disorder, DID, Mean number of comorbid Axis I Nether- 29.45 7-11 23.26% aged 18 to 60 ASPD, ADHD, diagnoses in the SFT group was 2.95 lands (6.5) Housewife 15.12%; years, BPDSO- addiction of (SD=0.23) and 2.40 (SD=0.25) in the student 10.47%; IV score >20, severity indicating TFP group. Mean number of employed 19.77%; Dutch literacy detoxification, comorbid Axis II diagnoses was 2.14 disability 39.53%; psychiatric (SD=0.18) in the SFT group and 2.05 welfare 15.12% disorders (SD=0.18) in the TFP group. Marital status not secondary to reported. medical conditions, mental retardation Jergen- 85 MBT = 95% Not reported. MBT group years of Adults > 20 Diagnosis of ASPD Participants were referred from a sen 30 (6.5) education < 15 0%, 13- fulfilled DSM- or paranoid range of outpatient and inpatient (2012) 15 years 19%, 10-12 IV criteria for personality treating facilities. SGT = years 33%, <10 years BPD, GAF disorder, severe Denmark 30 (6.8) 48% score >34 substance abuse Employed 10%, student Mean number of comorbid Axis I 14%, social security 69%, diagnoses was 2.0 (SD=1.3) in the pension 2% MBT group and 1.3 (SD=1.0) in the Married/partner 31%, SGT group. 72% of participants in successive partners 21%, the MBT group and 85% in the SGT single 48% group had at least one other personality disorder (other than SGT group years of BPD). education < 15 7%, 13- 15 years 7%, 10-12 years 52%, <10 years 33% Employed 7%, student 19%, social security 56%, pension 15% 19 Married/partner 41%, successive partners 22%, single 37% Gratz 22 ER + 100% White 100% Graduate school 31.82%; Female, 18 to Diagnosis of a Participants were referred from (2006) TAU = college graduate 36.36%; 60 years, psychotic disorder, local private practices, from 33.0 some college 18.18% meeting >5 bipolar I disorder, clinicians working in a local hospital, USA (12.47) criteria for and/or substance or self-referred. Income >$50 000 68.18% BPD in dependency, TAU = accordance reporting > 1 high- 59.10% of participants had been 33.70 Single 68.18% with DSM-IV, risk suicide hospitalized within the past year; (12.56) scoring > 8 on attempt within the 18.18% had attempted suicide the Revised previous 6 months, within the past year; 54.55% had Diagnostic reporting high-risk attempted suicide at least once Interview for of suicide attempt during their lifetime. Borderlines, within next 5 history of months, DBT skills deliberate group participation self-harm, within previous 6 having a months therapist, Gratz 61 33.32 100% Racial /ethnic College graduate 21.3%; DSM-IV Diagnoses of a All participants were women who (2014) (9.98) minority 16% High school graduate threshold or primary psychotic either self-referred or were referred 63.9%; Less than high subthreshold disorder, bipolar I by a treating clinician due to USA school 6.6%. diagnosis of disorder and difficulties with emotion regulation BPD (88.5% of current (past and self-harm. 19.7% of participants Income 20 Cluster B 16.39%, Cluster C 39.34% Koons 28 DBT = 100% Caucasian 75%; Bachelor’s degree or Female Diagnosis of Participants were recruited through (2001) 34.5 African equivalent 20%; some veterans who schizophrenia, the Women Veterans (7.5) American 25% college education 80% met criteria bipolar disorder, Comprehensive Health Centre in USA for BPD in substance North Carolina, as well as through TAU = Income < $20 000 accordance dependence, and other counselling centres 35.4 with antisocial throughout the state. All (6.9) Marital status not DSM-III-TR. personality participants had at least 1 reported. disorder. psychiatric outpatient visit in the previous 6 months; 25% had an inpatient psychiatric admission in the last 6 months. Comorbid diagnosis included: substance use disorder 25%. Linehan 44 Not 100% Not reported. Not reported. Female, 18 to Diagnosis of Participants were referred by (1991) report- 45 years, > 7 schizophrenia, treating clinicians. ed. on the DIB bipolar disorder, USA and met DSM- substance Comorbidity statistics not reported. III criteria for dependence, or BPD, at least 2 mental retardation incidents of parasuicide in the last 5 years (1 within the last 8 weeks) Linehan 28 30.4 100% European Some college/college Female, 18 to Met criteria for Participants were referred by local (1999) (6.6) decent 87%; graduate 63%; high 45 years, met schizophrenia, clinicians. African- school graduate or GED criteria for other psychotic USA American 7%; 22% BPD on both disorder, bipolar Comorbidity for Major Depressive Latina 4%; the PDE and mood disorder, or Disorder was 55% DBT, 45% TAU; Other 11% Income >$20,000 12%; the SCID-II, mental retardation Dysthymia 36% DBT, 55% TAU; Panic 21 $5,000-$19,999 35%; met criteria disorder 36% DBT, 0% TAU; <$5,000 54% for Substance agoraphobia 0% DBT, 9% TAU; Social Use Disorder Phobia 9% DBT, 36% TAU; Specific Single 63%; Married Phobia 30% DBT, 9% TAU; 15%; Separated 7%; Obsessive-Compulsive Disorder 30% Divorce 15% DBT, 27% TAU; PTSD 50% DBT, 27% TAU; Generalised Anxiety Disorder 40% DBT, 9% TAU; Anorexia Nervosa 0% DBT, 9% TAU; Bulimia Nervosa 10% DBT, 10% TAU; Binge-Eating Disorder 20% DBT, 0% TAU Linehan 23 36.1 100% Caucasian 66%; College graduate 4%; Female, 18 to Diagnosis of All participants were recruited from (2002) (7.3) African- some 45 years, bipolar disorder, mental health clinics and substance American 26%; graduate/professional diagnosis of psychosis, seizure abuse clinics within the community. USA Other 4% school 18%; some BPD on PDE disorder, or college or technical and SCID-II for mental retardation The mean number of comorbid Axis school 48%; high school DSM-IV, I diagnoses was 2.4 (SD=1.3); 39% graduate or general diagnosis of Major Depressive Disorder or equivalent 96% current opiate Dysthymia; 52% current anxiety dependence disorder; 18% eating disorder. Only Employed 52% according to anti-social personality disorder was the SCID-I, assessment for Axis II comorbidity; Divorced 52%l Married absence of 44% anti-social personality disorder 4%l never been married pregnancy or 44% any medical condition in which the use of opiate- replacement education was contra- indicated, absence of indications of 22 treatment coercion Linehan 101 29.3 100% African College graduate 23.8%; Female, aged Lifetime diagnosis All participants were clinically (2006) (7.5) American 4.0%; some college or 18 to 45 years, of schizophrenia, referred prior to being assessed for Asian American technical school 51.5%; met criteria schizoaffective participation eligibility. USA 2.0%; Native High school graduate or for BPD in disorder, bipolar American or general equivalency accordance disorder, psychotic Alaskan Native 16.8%l < high school with DSM-IV, disorder NOS, or Current comorbid diagnoses: major 1.0%; White 7.9% current and mental depressive disorder 72.3%; panic 87.0%; Other past suicidal retardation, a disorder 40.6%; PTSD 49.5%; any 5.0% Income >30 000-50 000 behaviour seizure disorder anxiety disorder 78.2%; any 9.9%; 15 000-30 000 defined by >2 requiring substance use disorder 29.7%; any 10.9%; <15 000 75.2% suicide medication, a eating disorder 23.8%. Comorbidity attempts or mandate to rates for Axis II disorders: Cluster A Single, divorced or self-injuries in treatment, or need 3.0% (Paranoid PD 3.0%); Cluster B separated 87.2% the past 4 for primary 10.9% (Antisocial 10.9%, Histrionic years, with at treatment for 2.0%), Cluster C 25.7% (Avoidant PD least 1 in the other debilitating 20.8%; Dependent PD 5.9%; past 8 weeks condition Obsessive-Compulsive PD 7.9%) McMain 180 30.36 86% Not reported. College graduate 16.7%; Adults aged Diagnosis of Current comorbid DSM-IV (2009) (9.9) some college education 18 to 60 years, psychotic disorder, diagnoses: major depressive 23.3%; high school meeting bipolar I disorder, disorder 48.9%; panic disorder Canada graduate 28.9%; 23 Annual income: <$15 one within 3 requiring 000 61.8%; $15 000 to months of hospitalisation $29 000 21.9%; $30 000 study within the next yr, to $49 000 9.6%; >$50 participation plans to leave area 000 6.9% in next 2 years Verheul 58 34.9 100% Not reported. Average years in Female, met DSM-IV diagnosis Participants referred from an (2003) (7.7) education 13.1 (SD=3.6) criteria for of bipolar disorder unrestricted range of referral BPD, aged 18 or (chronic) sources, including addition services, The Married 36% to 70 years, psychotic disorder, psychiatric hospitals, mental health Nether- living within a insufficient Dutch care centers, private practitioners, lands Unemployed 12%; 40km radius language general practitioners and self- disability pension 34% of proficiency, severe referral. Amsterdam, cognitive with a treating impairments Comorbid diagnoses not reported. psychologist/p sychiatrist willing to sign agreement to deliver 12 mths of TAU Table 1.2 Intervention Descriptions of Efficacy RCTs of Treatment of Borderline Personality Disorder (BPD) in Adults Authors Intervention Summary Supervision Session frequency / length Summary of control / Attendance Study Consultation Duration active comparison Design Adherence Delivered by Bateman MBT is a psychodynamic Adherence Session frequency/length: MBT involved 1x TAU involved general Average length of (1999) treatment informed by monitored weekly individual (60 mins), 3x weekly psychiatric services, stay in MBT was attachment and cognitive through twice group sessions (60 mins each), 1x weekly including a range of 1.45 years, and RCT theory, involving both weekly expressive therapy (60 mins), 1x weekly inpatient and outpatient attendance at the MBT individual and group supervision, using community meeting (60 mins), 1x monthly community treatment program’s (n=19) vs psychotherapy components. verbatim session case administrator meeting (60 mins) and options. No formal psychotherapy TAU Treatment is manualised report and medication review. Duration: 18 months. psychotherapy provided. sessions was 62%. (n=19) and delivered in an completion of Delivered by: psychiatrically trained nurses 24 outpatient context. monitoring form collecting Session frequency/length: TAU involved information twice-monthly, on average, psychiatric about activities review, outpatient community care and interventions (approximately every 2 weeks). Duration: 18 of therapists months. Delivered by: a range of clinicians (inpatient and outpatient as required) Bateman See above - Bateman (1999) Adherence Session frequency/length: MBT involved SCM was protocol-driven. MBT: average (2009) monitored weekly combined individual and group It was developed and number of through twice sessions. Duration: 18 months. Delivered informed by best practice scheduled clinical RCT MBT weekly by: two therapists guidelines for treatment of meetings attended (n=71) vs supervision, using BPD, involving individual was 92 (SD=38) SCM session report & Session frequency/length: SCM involved and group supportive (n=63) monitoring form regular individual and group sessions counselling, case SCM: average collecting (frequency unspecified), psychiatric review management, advocacy number of information every 3 months. Duration: 18 months. support, and problem- scheduled clinical about activities Delivered by: unreported. oriented meetings attended and interventions psychotherapeutic was 84 (SD=40) of therapists. interventions. Blum STEPPS + TAU involved Treatment Session frequency/length: STEPPS involved TAU group were Those allocated to (2008) group outpatient treatment. delivered by one weekly sessions which lasted 120 mins in encouraged to continue the STEPPS group Participant attended face- of the treatment duration. Duration: 20 weeks. Delivered by: with usual psychiatric care treatment attended RCT to-face group sessions and developers and one of the treatment developers and one of (e.g., individual an average of 12.9 STEPPS completed weekly one of the co- the co-authors (two highly-trained co- psychotherapy, (SD=5.4) sessions. Group + homework tasks. Primary authors of the facilitators) medication, and case TAU treatment components study. management). (n=65) vs include psychoeducation, TAU and skills development of Adherence (n=59) cognitive, emotional and measured and behavioural capacities. met. Family members, significant others, health care professionals also educated 25 about BPD and how to best support their relative/friend/patient. Bos Dutch version of STEPPS. A Therapists met Session frequency/length: STEPPS involved TAU group were offered STEPPS: average (2010) group program with three twice yearly weekly group sessions and bi-weekly standard treatment at the number of group key components (1) under supervision individual sessions. Duration: 18 weeks. participating sites sessions attended RCT psychoeducation about of expert trainers. Delivered by: two mental health (individual therapy, DBT or was 15. Median STEPPS BPD, (2) emotion Individual professionals (one a psychotherapist) STEPPS-related treatments number of (n=42) vs management skills training, therapists were not allowed). individual sessions TAU (3) behaviour management received a 1-day Session frequency/length: TAU involved attended: (n=37) skills training. Friends and training and individual sessions offered every 1 to 4 STEPPS = 16 family of the patient were monthly phone weeks (duration unspecified). Duration: 18 TAU = 18 explicitly involved for supervision. weeks. Delivered by: a psychologist, support. Following each psychotherapist, or psychiatric nurse session, self- report questionnaire used to check content of therapy. Clarkin TFP is a psychodynamic Supervised by a Session frequency/length: TFP involved two DBT group involved Not reported. (2007) therapy for patients with treatment individual weekly sessions. Duration: 52 individual and group BPD focusing on the condition leader weeks. Delivered by: administered and sessions in addition to RCT integration of internalized who was an supervised by a treatment condition leader telephone consultation as TFP experiences of early acknowledged and other selected therapists required. Content focused (n=23) vs dysfunctional relationships. expert. on emotion regulation DBT TFP involved individual Session frequency/length: DBT weekly skills. (n=17) vs treatment sessions focused Weekly feedback involved and group sessions. Duration: 52 supportiv on affect-laden themes provided to weeks. Delivered by: administered and ST involved individual e arising within the therapists by the supervised by a treatment condition leader sessions whereby general treatmen therapeutic relationship. condition leader and other selected therapists emotional support was t (ST) Limit setting, impulsivity provided (e.g., problem (n=22) reduction, interpersonal Session frequency/length: ST involved one solving daily problems). skills, daily functioning and individual weekly session. Duration: 52 26 mood stabilization are also weeks. Delivered by: administered and importance therapeutic supervised by a treatment condition leader components. and other selected therapists Cottraux CT (Cognitive Therapy) is a Therapists Session frequency/length: CT involved one Rogerian Therapy (RST) is The average (2009) directive therapy aiming to received individual weekly session for 6mths (60 min non-directive and includes amount of time challenge maladaptive supervision in CT duration), and one individual fortnightly therapeutic alliance and before ending RCT CT cognitive schemas that for BPD by an session for the following 6mths (60 min specified humanistic therapy was 82 (n=33) vs. relate to destructive experienced duration). Patients could contact their supportive techniques of days for CT and 60 RST emotions and behaviours, clinician. RST therapist or the principal investigators in psychotherapy. Treatment days for RST (n=32) both intra- and principles and the event of a crisis during working hours targets intra-and (p=0.13). interpersonal. Treatment methods taught (emergency department contacts provided interpersonal attitudes, Participants who targets a range of processes, through 10hr role for out-of-hours crises). Duration: 1 year. emotional expression, self- dropped out of including cognitive play. Clinicians Delivered by: Therapists training in both CT esteem and personal therapy left later in challenging, emotional from both and RST, the therapists were the same in growth. CT (M= 51 days; wellbeing, impulsive treating sites met both treatment groups. Treatment was SD= 37.4) behaviours, and self- for 12hrs to manualized. compared to RST regulation. consolidate Session frequency/length: RST involved one (M=29 days; Treatment was manualized. treatment individual weekly session for 6mths (60 min SD=32.4), p=0.04. duration), and one individual fortnightly techniques and session for the following 6mths (60 min discuss measures. duration). Patients could contact their therapist or the principal investigators in Therapists the event of a crisis during working hours completed post- (emergency department contacts provided session checklists for out-of-hours crises). Duration: 1 year. to identify Delivered by: Therapists training in both CT treatment and RST, the therapists were the same in techniques, later both treatment groups. discussing with PI. Weekly supervision was provided. 27 Davidson CBT + TAU involved a Therapists had 3 Session frequency/length: 30 individual TAU involved standard The average (2006) structured intervention days of training sessions (60 min duration) across a 12mth treatment interventions number of CBT tailored for individuals with prior to trial & 4 period. Duration: 1 year. Delivered by: 5 that were received sessions attended RCT CBT BPD. Treatment involves the days during trial. therapists (4 registered nurses, 1 irrespective of the was 16 (SD=12). + TAU discussion of a collaborative occupational therapist). 3 therapists had treatment trial. Treatment Attendance for TAU (n=54) vs. formulation of the patient’s Received completed a 10-month CBT training course, depended on the specific not reported. TAU difficulties, as well as fortnightly 1 therapist had received CBT training in problems of the individual. (n=52) implementing strategies to supervision. psychosis, 1 therapist had no CBT training manage and challenge but held experience managing individuals maladaptive beliefs and With patient with personality disorders. behaviours that impair daily consent, samples functioning. Priority during of sessions were treatment is given to audiotaped and reducing and self-harm evaluated for behaviours. adherence (using tailored version of The Cognitive Therapy Rating Scale). Doering See above – Clarkin (2007) Video recordings Session frequency/length: TFP involved two CP involved individual Average number of (2010) of sessions were weekly sessions lasting 50 mins in duration. psychotherapy a sessions for TFP done & used in Duration: 1 year. Delivered by: 31 therapists administered by was 48.5 (SD=34.2), RCT group supervision providing individual TFP community and 18.6 (SD=24.0) TFP (avg 2 hrs per psychotherapists, for CP. (n=52) vs. week for TFP & Session frequency/length: Community treatment modality CP (n=52) according to usual psychotherapy (CP) sessions unspecified. practice for CP). frequency/length determined by individual TFP therapists psychotherapists. Duration: 1 year. received more Delivered by: community psychotherapists supervision. Assessment of adherence was done for TFP. 28 Farrell SFT involved group therapy Two of the three Session frequency/length: SFT involved TAU group participated in Not reported. (2009) content focusing on four key groups had the 2 weekly group sessions which lasted 90 mins individual treatment-as- components including program in duration. Duration: 30 weeks. Delivered usual alone. Participants in RFT SFT + emotional awareness developers as by: two active therapists the TAU group were TAU vs. training, BPD therapists and the required to attend weekly TAU psychoeducation, distress third had one of individual psychotherapy management training, and the program as a condition of schema change work. developers. remaining in the study. Schema change components Weekly incorporated element of supervision schema therapy adapted for sessions. individuals with BPD. Giesen- SFT involved a focus on Weekly Session frequency/length: SFT involved TFP aims to achieve SFT participants Bloo treating supervision in twice-weekly individual sessions lasting 50 positive change through attended (2006) maladaptive/dysfunctional both models; 1 mins in duration. Duration: 3 years. close analyses of the significantly fewer schema modes considered day central Delivered by: therapists with a range of therapeutic relationship sessions that those RCT SFT specific to BPD. Change is supervision every qualifications (1 doctoral degree, 19 and subsequent completing TFP (n=44) vs. achieved through cognitive, 4 mths and 2 day master’s degrees, 3 bachelor’s degrees with transference, through the (median sessions TFP behavioural, and central postgraduate training), all with experience use of exploration, SFT=189.6, (n=42) experiential treatment supervision every treating patients with BPD TFP=231.0). techniques tending to: the 9 months by confrontation, and therapeutic relationship, treatment expert. Session frequency/length: TFP involved interpretation. daily life, past traumatic twice-weekly individual sessions lasting 50 experiences. Audiotapes were mins in duration. Duration: 3 years. evaluated for Delivered by: therapists with a range of integrity. qualifications (2 doctoral degrees, 18 master’s degrees), 1 bachelor degree with postgraduate training, all with experience treating patients with BPD Gratz ERGT is an acceptance- Initial training of Session frequency/length: ERGT involved TAU involved continuing Average number of (2006) based, behavioural group therapists lasted weekly group session (90 mins duration) in with current outpatient hours/week spend based on a approx. 4 months addition to TAU. Duration: 14 weeks. treatment over the course in therapy was 2.10 Pilot RCT multidimensional with ongoing Delivered by: not reported. of the study. Thirty-two (SD=1.56) for ERGT ERGT + conceptualization of supervision percent of TAU group, and 2.95 29 TAU emotion regulation. It throughout the Session frequency/length: TAU outpatient participants also attended (SD=2.78) for TAU (n=12) vs. involved targeting emotion trial. PI reviewed treatment differed in frequency and length. group therapy and 9% group. TAU regulation skills and self- all group sessions Duration: 14 weeks. Delivered by: clinical attended self-help groups. (n=10) harm. Participants for adherence psychologists, psychiatrists, clinical social continued with ongoing and competence workers outpatient therapy during (25% were rated ERGT. Content of ERGT was by independent informed by Acceptance and rater). Commitment Therapy & DBT. Gratz See above – Gratz (2006). See above – Gratz Session frequency/length: ERGT involved TAU group continued with Not reported. (2014) (2006). weekly sessions which lasted 90 mins in outpatient therapy whilst duration. Duration: 14 weeks. Delivered by: placed on waitlist for RCT Two doctoral-level therapists. ERGT. Group + TAU group later TAU completed ERGT following (n=31) vs the end of the waitlist TAU period. (n=30) Jer- Combined MBT is informed Supervision for 2 Session frequency/length: MBT involved SGT focused on each Not reported. gensen by mentalisation theory and hours per week weekly individual sessions (45 mins participants’ subjective (2012) attachment theory. using video duration), weekly group sessions (90 mins experiences in and outside Individual and group recordings of duration), and psycho-educational groups of group therapy. RCT sessions focused on the their individual (once/months for 6 months). Duration: 2 Strategies included Combine therapeutic relationship as therapy sessions. years. Delivered by: six experienced problem-solving, d MBT well as patient’s interpersonal therapists with substantial managing conflict, and (n=58) vs. relationships with one psychodynamic training behavioural strategies to SGT another. Treatment aimed build adaptive responses. (n=27) to develop the patient’s Session frequency/length: Supportive group ability to mentalise and therapy (SGT) involved biweekly groups develop more adaptive sessions (90 mins duration) and psycho- interpersonal behaviours. educational groups (once/month for 6 months). Durations: 2 years. Delivered by: 30 two experienced therapists Koons DBT involved the Weekly Session frequency/length: DBT involved TAU implemented type of Not reported. (2001) implementation of both consultation weekly individual sessions (length not therapy they deemed best individual and group group and reported) and group skills training sessions for each participant with a RCT treatment targeting suicidal additional (90 mins). Duration: 6 months. Delivered by: diagnosis of BPD. DBT and self-injurious supervision as 1 psychiatrist, 2 psychologists, a clinical Participants were also (n=10) vs behaviours through the use needed. social worker, and a clinical nurse specialist offered to attend one or TAU of behavioural therapy in psychiatry (first five authors). DBT groups more supportive and (n=10) strategies. Treatment was Sessions facilitated by clinical nurse specialist and a psychoeducation groups. based upon Linehan’s (1993) videotaped and psychiatry resident. treatment manual. coded for adherence. Session frequency/length: TAU involved weekly individual sessions which lasted 60 mins in duration. Duration: 6 months. Delivered by: 3 psychologists, 2 resident psychiatrists, 2 clinical social workers and a clinical nurse specialist in psychiatry (none with specialised DBT training). Linehan DBT involved a manualized DBT sessions Session frequency/length: DBT involved TAU participants were Not reported. (1991) individual and group supervised by weekly individual (60 mins duration) and provided with alternative RCT treatment combining Marsha Linehan group (150 mins duration) sessions, in therapy referrals from DBT cognitive, behavioural and who trained all addition to skills coaching phone calls when whom they could recent (n=22) vs. supportive therapeutic therapists, needed. Duration: 1 year. Delivered by: 5 treatment for BPD. TAU components. Behavioural listened to psychologists, 1 clinical psychology graduate (n=22) skills training is balanced by audiotapes at student, 1 psychiatrist acceptance strategies. regular intervals Overall focus is and conducted Session frequency/length: TAU session management of emotional weekly individual frequency and length varied. Duration: 1 trauma and building and group year. Delivered by: range of treating towards a more fulfilling life. supervision. clinicians Linehan DBT implemented as DBT sessions Session frequency/length: DBT involved TAU involved referral to Not reported. (1999) individual and group supervised by weekly individual (60 mins duration) and alternative substance sessions incorporating Marsha Linehan group (120 min duration) sessions, in abuse and/or mental RCT acceptance/validation with who trained all addition to skills coaching phone calls when health 31 DBT dedication to behavior therapists, and needed. Duration: 1 year. Delivered by: 2 counselors/community (n=12) vs. change. Modifications to conducted weekly psychologists, 1 psychiatrist, 2 master’s programs, or continue TAU standard DBT included a individual and level clinicians with TAU. (n=15) focus on attachment, group dialectical abstinence, and supervision. Session frequency/length: TAU session transitional maintenance of Sessions coded frequency and length varied. Duration: 1 replacement medications for adherence year. Delivered by: range of treating for participants with clinicians stimulant or opiate dependence. Linehan DBT involved directive, Weekly Session frequency/length: DBT involved Comprehensive validation DBT participants (2002) problem-oriented supervision to weekly individual sessions (40-90 mins) and therapy for substance attended avg of techniques balanced with review cases and weekly group sessions (150 mins), 12-Step abusers (CVT + 12S) is a 33.2 (SD=20.4) RCT DBT supportive techniques (e.g., session support groups recommended, case manualised treatment individual sessions, (n=11) vs. reflection, empathy, and videotapes. management and phone consultation as aiming to provide support, avg of 26.69 CVT + 12S acceptance). Linehan’s required. Duration: 1 year. Delivered by: 2 validation and general (SD=15.9) skills (n=12) treatment manuals were doctoral-level and 1 masters level behaviour therapeutic acceptance group sessions, avg adapted for substance therapists and other components of of 17.6 (SD=9.9) abusers. treatment not specific to coaching sessions DBT. An amalgamation of Session frequency/length: CVT + 12S DBT acceptance-based CVT participants involved weekly individual (40-90 mins) and strategies and attended avg of group sessions (120 min), 12-Step sponsor environmental 33.0 individual meetings recommended, 12-Step meetings intervention when (SD=9.6) sessions, recommended, CVT case management, and requested. avg of 10.8 phone consultation as required. Duration: 1 (SD=12.8) group year. Delivered by: 2 masters level meetings, avg of therapists with chemical-dependency 6.7 (SD=2.5) certification and 12-step experience. coaching sessions. Linehan See above – Linehan (1991) Initial DBT Session frequency/length: DBT involved 1 CTBE involved an The median (2006) training included weekly individual session (60 mins duration) alternative treatment-as- number of 45 hour seminar and one weekly group skills training session usual condition, whereby individual sessions RCT followed by (150 mins duration). Telephone expert clinicians provided during the study 32 DBT supervised consultation was provided as required. type and dose of therapy year for those (n=52) vs practice. Duration: 1 year. Delivered by: 3 graduate they believe most suited allocated to DBT CTBE students, 2 postdoctoral trainees, all the patient. Minimum was 42.5, and group Adherence was completed 45-hours of DBT training in requirements included 1 group DBT sessions (n=49) assessed with addition to receiving supervised practice. scheduled weekly session. was 38.0. random selection CTBE therapists selected of sessions. based on The median recommendations of number of CBTE therapists community mental health individual sessions attended weekly leaders. Content of during the study supervision. treatment not prescribed, year for those rather individual allocated to CTBE therapists provided the was 33.0, with no type and dose they saw group sessions. best fit each client. McMain DBT consisted of a Weekly individual Session frequency/length: DBT involved 1 General Psychiatric DBT condition: (2009) manualised cognitive- supervision plus weekly individual session (60 mins), 1weekly Management (GPM) was Mean number of behavioural treatment, approximately group skills training session (120 mins), and based on practice treatment weeks RCT aiming to eliminate monthly expert weekly phone consultation (120 mins). guidelines for the attended =36 behavioural dyscontrol by consultation Duration: 1 year. Delivered by:3 treatment of PBD (SD=17.57). Mean DBT teaching effective coping psychiatrists, 4 psychologists with Ph.D published by the APA and number of (n=90) vs skills. Adherence coding qualifications, 5 master’s level clinicians, manualised for the trial. individual session GPM conducted and and 1 nurse. All therapists had a minimum GPM included case attended = 32 (n=90) feedback of 1 year of experience treating BPD. management, dynamically SD=32.97). provided to informed psychotherapy, therapists Session frequency/length: GPM involved 1 and symptom-targeted GPM condition: weekly individual session (60 mins). medication management. Mean number of Duration: 1 year. Delivered by: 8 treatment weeks psychiatrists, 1 psychologist with Ph.D attended =27 qualifications, 1 master’s level clinician, and (SD=14.84). Mean 2 nurses. All therapists had a minimum of 1 number of year of experience treating BPD. individual session attended = 31 SD=27.05). 33 Verheul DBT involved manualised Training, regular Session frequency/length: DBT involved 1 TAU involved clinical Not reported. (2003) treatment that comprised a monitoring (using weekly individual session (60 mins) and 1 management from combination of individual videotapes) and weekly skills-training group session (120- participants’ current RCT DBT and group-based cognitive- weekly individual 150 mins). Duration: 1 year. Delivered by: 4 treating clinician (referral (n=27) vs behavioural psychotherapy and group psychiatrists and 12 clinical psychologists. source). TAU involved no TAU sessions including skills- supervision were Group sessions co-facilitated by social more than 2 sessions per (n=31) training, motivational issues, performed by a workers and clinical psychologists. month. self-regulation and self- DBT expert acceptance skills. Table 1.3 Summary of Results of Efficacy RCTs of Treatment of Borderline Personality Disorder (BPD) in Adults Authors Assess- Attrition (over study) Outcome measures and results Drop-out Study ment rates from Design points treatment (months) Bateman 3, 6, 9, 60 referred patients Outcomes: MBT = 12% (1999) 12, 15, met inclusion criteria, Suicide and self-harm (SSHI): Signif improvement in median number of self-harm incidents for 18 (post) 10 refused to MBT from 9 to 1 vs TAU group reduction from 8 to 6, p<0.01. TAU not RCT participate in random Signif improvement in proportion of suicide attempts for MBT (Baseline mean of 1.68 and 94.7% reported MBT allocation, 6 did not at Baseline to mean of 0.16 and 5.3% at 18 mths), vs TAU (proportions and means not reported (n=19) vs wish to participate in for TAU) p<0.05. TAU self-assessment. 44 General Psychiatric Symptoms (SCL-90-R): Signif improvement for MBT (Baseline Mean 2.50 (SD 34 (n=19) participants in the 0.58), 6mth 2.40 (SD 0.51), 12mth 2.20 (SD 0.60), 18mth 2.10 (SD 0.82), ES 0.69) vs TAU (Baseline study. 3 in TAU group Mean 2.30 (SD 0.71), 6mth 2.40 (SD 0.67), 12mth 2.40 (SD 0.69), 18mth 2.40 (SD 0.70), ES -0.14), transferred to MBT p<0.01. group in 1st month Depression (BDI): Signif improvement for MBT (Baseline Mean 26.0 (SD 7.6), 3mth 36.2 (SD 7.3), due to suicide 6mth 36.6 (SD 8.9), 9mth 30.7 (SD 10.4), 12mth 26.7 (SD 8.7), 15mth 23.7 (SD 5.7), 18mth 20.6 (SD attempts. 3 patients 7.0) ES 2.0) vs TAU (Baseline Mean 34.9 (SD 7.4), 3mth 35.0 (SD 6.5), 6mth 36.5 (SD 10.1), 9mth in MBT group 34.2 (SD 9.2), 12mth 34.7 (SD 9.1), 15mth 36.3 (SD 10.2), 18mth 35.2 (SD 7.4) ES 0.04), p<0.001. dropped out within 6 Trait Anxiety (STAI-T): Signif improvement for MBT (Baseline Mean 66.5 (SD 6.1), 3mth 65.8 (SD months. Final 5.8), 6mth 62.3 (SD 9.8), 9mth 60.3 (SD 7.4), 12mth 60.4 (SD 7.4), 15mth 56.4 (SD 8.9), 18mth 56.8 analysis included (SD 9.1) ES -1.59 vs TAU 62.0 (SD 9.9), 3mth 61.6 (SD 8.9), 6mth 62.5 (SD 5.2), 9mth 62.1 (SD 7.0), MBT (n=19) and TAU 12mth 60.6 (SD 7.0), 15mth 60.5 (SD 9.2), 18mth 61.0 (SD 7.6) ES -0.10), p<0.05. (n=19). Social Adjustment (SAS): End-of-treatment (18mth) SAS score significantly lower for MBT than TAU (Mean of 2.8 vs 3.3, p<0.01). Standard deviations not reported. Interpersonal Problems (IPP): Signif improvement for MBT (Baseline Mean 2.38 (SD 0.33), during treatment 1.86 (SD 0.36) vs TAU 2.31 (SD 0.32), during treatment 2.60 (SD 0.29)), with a highly signif difference between the groups, p<0.001). N.B. It was not stated which of the above were primary vs secondary outcomes. Bateman 3, 6, 12, 168 screened for Primary outcomes: MBT = 27% (2009) 18 (post) eligibility. 10 did not Hospital admissions, suicidal and self-harm episodes (medical records): Signif improvement for attend interview. 158 MBT (Baseline Mean 5.7 (SD 5.6), 6 mths 3.4 (SD 3.3), 12 mths 1.8 (SD 2.7), 18 mths 0.5 (SD 0.9) ES SCM = 25% RCT MBT participants 0.93) vs SCM (Baseline Mean 5.1 (SD 4.1), 6 mths 2.7 (SD 3.0), 12 mths 2.6 (SD 2.7), 18 mths 2.2 (n=71) vs interviewed, 5 did (SD 3.3) ES 0.71), p<0.001. SCM not meet inclusion Life-threatening suicide attempts (medical records): Signif improvement for MBT (Baseline Mean (n=63) criteria, 4 met 1.28 (SD 1.15), 6 mths 0.62 (SD 0.74), 12 mths 0.36 (SD 0.57), 18 mths 0.03 (SD 0.17) 1.09) vs SCM exclusion criteria, 3 (Baseline Mean 1.0 (SD 0.92), 6 mths 0.70 (SD 0.81), 12 mths 0.60 (SD 0.77), 18 mths 0.32 (SD unable to contact. 0.62) ES 0.74), p<0.001. 146 participants Severe self-harm incidents (medical records): Signif improvement for MBT (Baseline Mean 4.11 enrolled, 12 refused (SD 4.90), 6 mths 2.61 (SD 3.08), 12 mths 1.30 (SD 2.47), 18 mths 0.38 (SD 0.83) ES 0.76) vs SCM randomization. 134 (Baseline Mean 3.75 (SD 3.69). 6 mths 1.79 (SD 2.62), 12 mths 1.73 (SD 2.27), 18 mths 1.66 (SD entered treatment 2.86) ES 0.57), p<0.001. groups (MBT n=71, Psychiatric hospitalisation (medical records): Signif improvement for MBT (Baseline Mean 0.31 (SD SCM n=63) and were 0.55), 6 mths 0.13 (SD 0.33), 12 mths 0.08 (SD 0.28), 18 mths 0.03 (SD 0.17) ES 0.51) vs SCM included in analyses. (Baseline Mean 0.32 (SD 0.53), 6 mths 0.24 (SD 0.42), 12 mths 0.26 (SD 0.54), 18 mths 0.19 (SD 0.40) ES 0.25), p<0.001. 35 Length of hospitalisation (medical records): Signif improvement for MBT (Baseline Mean 5.41 (SD 11.66), 6 mths 0.28 (SD 0.78), 12 mths 0.70 (SD 2.64), 18 mths 0.20 (SD 1.18) ES 0.45) vs SCM (Baseline Mean 5.97 (SD 12.42), 6 mths 2.62 (SD 5.78), 12 mths 4.11 (SD 10.47), 18 mths 1.32 (SD 3.55) ES 0.37), p<0.001. Secondary outcomes: Functioning (GAF): Signif improvement for MBT (Baseline Mean 41.0 (SD 8.4), 18 mths 60.9 (SD 15.8) ES 2.37) vs SCM (Baseline Mean 41.0 (SD 8.4), 18 mths 53.2 (SD 11.7) ES 1.45), p<0.001. Symptom distress (SCL-90-R): Signif improvement for MBT (Baseline Mean 1.97 (SD 0.69), 6 mths 1.77 (SD 0.70), 12 mths 1.54 (SD 0.72), 18 mths 1.12 (SD 0.61) ES 1.23) vs SCM (Baseline Mean 2.02 (SD 0.60), 6 mths 1.91 (SD 0.65), 12 mths 1.81 (SD 0.68), 18 mths 1.55 (SD 0.66) ES 0.78), p<0.001. Depression (BDI): Signif improvement for MBT (Baseline Mean 29.83 (SD 10.09), 6 mth 26.19 (SD 9.64), 12 mth 20/59 (SD 9.51), 18 mth 14.80 (SD 8.55) ES 1.49) vs SCM (Baseline Mean 29.11 (SD 8.81), 6 mth 26.29 (SD 8.13), 12 mth 22.37 (SD 8.63), 18 mth 18.86 (SD 9.76) ES 1.18), p<0.01. Social Adjustment (SAS): Signif improvement for MBT (Baseline Mean 2.74 (SD 0.46), 6 mth 2.48 (SD 0.59), 12 mth 2.23 (SD 0.56), 18 mth 1.76 (SD 0.50) ES 2.13) vs SCM (Baseline Mean 2.70 (SD 0.64), 6 mth 2.62 (SD 0.66), 12 mth 2.46 (SD 0.67), 18 mth 2.17 (SD 0.64) ES 0.83), p<0.001. Interpersonal functioning (IIP): Signif improvement for MBT (Baseline Mean 2.01 (SD 0.54), 6 mth 1.86 (SD 0.55), 12 mth 1.61 (SD 0.51), 18 mth 1.28 (SD 0.13) ES 1.35) vs SCM (Baseline Mean 2.04 (SD 0.47), 6 mth 2.03 (SD 0.53), 12 mth 1.94 (SD 0.51), 18 mth 1.65 (SD 0.55) ES 0.83), p<0.001. Blum 1, 2, 3, 4, 172 participants Primary outcome: Rating Scale for BPD (ZRSB): Signif improvement for STEPPS (Baseline Mean STEPPS + (2008) 5 (post), assessed for 18.9 (SD 6.8), 5 mth 9.8 (SD 1.0 ES 1.34) vs TAU (Baseline Mean 17.3 (SD 7.0), 5 mth 13.4 (SD 1.0) TAU =31% 17 FU eligibility, 7 excluded. ES 0.56), p<.001. RCT 165 participants Secondary outcomes: Clinical Global Impression (CGI): Signif improvement for STEPPS (Baseline TAU = 14% STEPPS randomized to Mean 5.1 (SD 0.8), 5 mth 4.4 (SD 0.1), 17 mth 4.4 (SD 0.2) ES 0.88) vs TAU (Baseline Mean 4.9 (SD Group + treatment conditions 0.9), 5 mth 4.7 (SD 0.1), 17 mth 4.9 (SD 0.2) ES 0.22), p<.001. TAU (STEPPS n=93, TAU Functioning (GAF): Signif improvement for STEPPS (Baseline Mean 39.7 (SD 11.2), 5 mth 50.5 (SD (n=65) vs n=72). STEPPS group: 1.6), 17 mth 53.0 (SD 2.0) ES 0.96) vs TAU (Baseline Mean 39.6 (SD 11.4), 5 mth 43.5 (SD 1.6), 17 TAU 27 did not receive mth 47.1 (SD 2.7) ES 0.34), p<.001. (n=59) intervention, 1 Depression (BDI): Signif improvement for STEPPS (Baseline Mean 29.0 (SD 11.6), 5 mth 22.0 (SD excluded due to 2.0), 17 mth 24.0 (SD 1.8) ES 0.60) vs TAU (Baseline Mean 29.7 (SD 15.0), 5 mth 25.8 (SD 2.0), 17 primary neurological mth 23.4 (SD 3.8) ES 0.26), p<.05. disorder, 65 included General Psychiatric Symptoms (SCL-90-R): Signif improvement for STEPPS (Baseline Mean 16.0 (SD in analysis, 45 7.2), 5 mth 12.5 (SD 1.0) ES 0.48) vs TAU (Baseline Mean 16.8 (SD 6.0), 5 mth 14.9 (SD 1.1) ES 36 completed 0.26), p<.05. intervention & Impulsiveness (BIS): Signif improvement for STEPPS (Baseline Mean 80.6 (SD 12.6), 5 mth 72.7 (SD assessment. TAU 1.8) ES 0.63) vs TAU (Baseline Mean 77.4 (SD 12.8), 5 mth 76.8 (SD 1.8) ES 0.05), p<.01. group: 13 did not Social Adjustment (SAS): No signif difference between STEPPS (Baseline Mean 27.8 (SD5.0), 5 mth receive intervention, 24.6 (SD 0.8), 17 mth 24.3 (SD 1.0) ES 0.64) vs TAU (Baseline Mean 28.2 (SD 5.0), 5 mth 26.3 (SD 59 included in 0.8), 17 mth 27.1 (SD 1.5) ES 0.38), ns. analyses, 51 completed intervention & assessment. Bos 4 (post), 198 patients Outcomes reported in published paper were treatment completer, not intention to treat sample. STEPPS = (2010) 10 FU assessed, 115 Outcomes: 21% excluded. 83 BPD Criteria (BPD-40): Signif improvement for STEPPS (Baseline Mean 99.4 (SD 25.8), 4 mths 75.9 RCT participants (SD 24.9), 10 mths 74.1 (SD 25.3) ES 1.10) vs TAU (Baseline Mean 92.7 (SD 31.7), 4 mths 85.8 (SD TAU = 11% STEPPS randomized, 4 32.6), 10 mths FU 81.5 (SD 35.6) ES 0.18), p<.01. (n=42) vs withdrew following General Psychiatric Symptoms (SCL-90): Signif improvement for STEPPS (Baseline Mean 249.8 (SD TAU randomization. 79 67.1), 4 mths 188.5 (SD 55.6), 10 mths 188.0 (SD 58.1) ES 0.82) vs TAU (Baseline Mean 239.6 (SD (n=37) participants allocated 72.5), 4 mths 233.4 (SD 81.5), 10 mths 216.7 (SD 84.2) ES -0.02), p<.01. to treatment (STEPPS n=42, TAU n=37). 33 of each completed Quality of Life (WHOQOL-Bref): Signif improvement for STEPPS (Baseline Mean 10.9 (SD 3.1), 4 post assessments. mths 13.4 (SD 2.9), 10 mths 13.5 (SD 3.50.31)) vs TAU (Baseline Mean 10.8 (SD 3.5), 4 mths 11.4 (SD 3.6). 10 mths 11.7 (SD 3.6) ES 0.31), p<.01. N.B. It was not stated which of the above were primary vs secondary outcomes. Clarkin 12 (post), 336 patients Outcomes: TFP = 23% (2007) 18 (FU), referred, 129 Suicidality (OAS): TFP and DBT significantly associated with improvement in suicidality over time 24 (FU) excluded. 207 (means and standard deviations not reported). DBT = 43% RCT TFP assessed, 109 eligible Anger (AIAQ): TFP and DBT significantly associated with reduction in anger over time (means and (n=23) vs for randomization, 90 standard deviations not reported). Supportive = DBT randomized. Depression (BDI): TFP, DBT and Supportive treatments saw significant improvements for 27% (n=17) vs Analyses completed depression. supportiv for those who Anxiety (BSI): TFP, DBT and Supportive treatments saw significant improvements for anxiety. e provided data at 3 Functioning (GAF): TFP, DBT and Supportive treatments saw significant improvements for global treatmen time points (TFP functioning. 37 t (ST) n=23, DBT n=17, STP Impulsivity (BIS): TFP saw significant reductions in Barratt Factor 2 of impulsivity, irritability, verbal (n=22) n=22). and direct assault. Supportive treatment predicted improvement in Barratt Factor 3 (impulsivity). N.B. It was not stated which of the above were primary vs secondary outcomes. Cottraux 6, 88 patients assessed, Outcomes: Not (2009) 12 (post), 13 excluded, 10 Clinical Global Impression (CGI) severity: No signif difference between CT (Baseline Mean 5.16 reported. 24 (FU) refused to continue. (0.93), 6mth 3.96 (1.25), 12mth 3.35 (1.42), 24mth 2.20 (1.03), ES 1.94 vs RST (Baseline Mean 5.26 RCT CT 65 randomised to (1.06), 6mth 4.56 (1.47), 12mth 3.78 (1.35), 24mth 3.18 (1.40), ES 1.39), ns (n=33) vs. treatment (CT=33; Depression (HAM-D): No signif difference between CT (Baseline Mean 17.74 (6.79), 6mth 11.04 RST RST=32). 51 patients (6.72), 12mth 10.65 (9.71), 24mth 6.40 (4.48), ES 1.04 vs RST (Baseline Mean 22.23 (8.07), 6mth (n=32) were re-evaluated at 16.0 (9.68), 12mth 9.89 (8.93), 24mth 11.55 (5.41), ES 1.53), ns 6mths, 38 at 12mths, Depression (BDI): No signif difference between CT (Baseline Mean 21.79 (8.26), 6mth 13.04 (8.08), and 21 at 24mths. 12mth 13.06 (11.29), 24mth 8.20 (7.08), ES 1.06 vs RST (Baseline Mean 26.35 (10.45), 6mth 21.72 (12.63), 12mth 12.56 (9.56), 24mth 16.18 (12.63), ES 1.32), ns Anxiety (BAI): No signif difference between CT (Baseline Mean 23.45 (14.03), 6mth 13.73 (9.31), 12mth 12.50 (7.80), 24mth 9.91 (6.11), ES 0.78 vs RST (Baseline Mean 23.94 (10.29), 6mth 18.20 (10.95), 12mth 18.44 (14.25), 24mth 21.73 (15.54), ES 0.53), ns Hopelessness (BHS): No signif difference between CT (Baseline Mean 10.68 (0.04), 6mth 7.54 (5.06), 12mth 7.25 (5.37), 24mth 5.30 (3.80), ES 85.75 vs RST (Baseline Mean 10.03 (4.97), 6mth 9.64 (5.51), 12mth 6.56 (4.98), 24mth 7.73 (4.84), ES 0.79), ns Schemas (YSQ-II): No signif difference between CT (Baseline Mean 647.23 (110.22), 12mth 513.75 (115.05), 24mth 431.82 (90.81), ES 1.21 vs RST (Baseline Mean 690.45 (149.71), 12mth 532.56 (125.03), 24mth 599.09 (193.89), ES 1.05), ns Venturesomeness (IVE): No signif difference between CT (Baseline Mean 6.74 (3.72), 6mth 7.65 (3.89), 12mth 7.30 (3.56), 24mth 6.64 (3.53), ES 0.15 vs RST (Baseline Mean 6.97 (3.38), 6mth 5.96 (3.78), 12mth 5.61 (3.60), 24mth 6.64 (4.00), ES 0.40), ns Impulsivity (IVE): No signif difference between CT (Baseline Mean 10.35 (4.57), 6mth 8.25 (4.82), 12mth 7.60 (3.82), 24mth 4.55 (4.66), ES 0.61 vs RST (Baseline Mean 11.45 (3.90), 6mth 9.80 (5.23), 12mth 8.61 (4.97), 24mth 6.73 (4.00), ES 0.73), ns Empathy (IVE): No signif difference between CT (Baseline Mean 15.10 (2.45), 6mth 14.92 (2.65), 12mth 15.00 (2.47), 24mth 14.55 (2.70), ES 0.04 vs RST (Baseline Mean 14.81 (2.46), 6mth 14.08 (3.04), 12mth 14.67 (2.66), 24mth 14.73 (2.90), ES 0.06), ns Self-harm (interview): No signif difference between CT (Baseline 4.39 (2.06), 6mth 1.96 (1.25), 12mth 1.80 (1.36), 24mth 0.73 (1.19), ES 1.26 vs RST (Baseline Mean 4.47 (2.16), 6mth 2.36 (1.6), 38 12mth 1.06 (1.06), 24mth 1.36 (1.43), ES 1.58), ns Quality of life (SAS): No signif difference between CT (Baseline Mean 24.71 (9.53), 6mth 15.88 (9.12), 12mth 12.85 (9.22), 24mth 11.80 (11.29), ES 1.24 vs RST (Baseline Mean 23.26 (8.58), 6mth 18.56 (11.18), 12mth 14.33 (9.31), 24mth 18.73 (10.22), ES 1.04), ns N.B. It was not stated which of the above were primary vs secondary outcomes. Davidson 12 (post), 110 patients referred Primary outcomes: Not (2006) 24 (FU), and assessed, 4 Suicidal acts (ADSHI): No signif difference between CBT + TAU vs TAU: Global odds ratio of 0.77 at reported. 72 (FU) excluded. 106 12 mths and 0.78 at 24 mths. ns RCT CBT randomised to No. of inpatient hospitalisations (medical records): No signif difference between CBT + TAU vs + TAU treatment conditions TAU: Global odds ratio of 0.79 at 12 mths and 0.63 at 24 mths. ns (n=54) vs. (CBT+TAU=54; No. of accident/emergency contacts (medical records): No signif difference between CBT + TAU vs TAU TAU=52). 5 lost to FU TAU: Global odds ratio of 2.05 at 12 mths and 1.81 at 24 mths. ns (n=52) at 6mths (1 due to Secondary outcomes: natural death), 1 lost Psychiatric Symptoms (BSI-GSI): No signif difference between CBT+TAU (Baseline Mean 2.6 (0.6), to FU at 12mths, 5 12mth 1.97 (0.91), 24mth 1.81 (1.00), 72mth 1.75 (1.02) ES 1.05 vs TAU (Baseline Mean 2.4 (0.9), lost to FU at 18mth, 5 12mth 2.00 (0.93), 24mth 1.93 ( 1.00), 72mth 1.68 (1.09) ES 0.44), ns lost to FU at 24mth. Depression (BDI-II): No signif difference between CBT+TAU (Baseline Mean 42.6 (10.1), 12mth 6-year FU data 29.6 (14.8), 24mth 26.5 (15.3), 72mth 26.5 (16.5) ES 1.29 vs TAU (Baseline Mean 42.5 (12.3), obtained for 82% of 12mth 31.3 (16.6), 24mth 28.8 (15.7), 72mth 26.2 (17.5) ES 0.91), ns. patients. Social Functioning Questionnaire (SFQ): No signif difference between CBT+TAU (Baseline Mean 14.9 (4.1), 12mth 13.1 (4.4), 24mth 13.0 (5.0), 72mth 10.3 (4.98) ES 0.44 vs TAU (Baseline Mean 14.3 (4.1), 12mth 13.1 (4.6), 24mth 12.3 (5.3), 72mth 11.2 (5.00) ES 0.29), ns. Interpersonal Problems (IIP): No signif difference between CBT+TAU (Baseline Mean 72.4 (16.0), 12mth 60.4 (23.9), 24mth 54.0 (23.9), 72mth 53.1 (23.7) ES 0.75 vs TAU (Baseline Mean 65.9 (17.4), 12mth 55.0 (22.3), 24mth 53.7 (24.1), 72mth 48.7 (24.6) ES 0.63), ns. State Anxiety (STAI-S): No signif difference between CBT+TAU (Baseline Mean 53.6 (12.2), 12mth 49.2 (14.8), 24mth 48.2 (14.4), 72mth 45.7 (13.5) ES 0.36 vs TAU (Baseline Mean 51.4 (12.0), 12mth 49.7 (15.5), 24mth 50.9 (15.7), 72mth 48.2 (13.0) ES 0.14), ns. Trait Anxiety (STAI-T): No signif difference between CBT+TAU (Baseline Mean 65.8 (7.8), 12mth 59.7 (10.3), 24mth 56.4 (11.9), 72mth 55.1 (13.3) ES 0.78 vs TAU (Baseline Mean 64.0 (8.6), 12mth 60.0 (11.2), 24mth 58.0 (10.9), 72mth 56.5 (13.6) ES 0.47), ns. Schemas (YSQ): No signif difference between CBT+TAU (Baseline Mean 4.13 (0.66), 12mth 3.49 (0.84), 24mth 3.46 (0.99), 72mth 3.09 (1.03) ES 0.97 vs TAU (Baseline Mean 3.78 (0.70), 12mth 39 3.44 (0.91), 24mth 3.48 (0.91), 72mth 3.16 (1.12) ES 0.49), ns. Quality of Life (EQ-5D): No signif difference between CBT+TAU (Baseline Mean 0.49 (0.37), 12mth 0.51 (0.41), 24mth 0.58 (0.36), 72mth 0.52 (0.38) ES 0.05 vs TAU (Baseline Mean 0.52 (0.36), 12mth 0.62 (0.35), 24mth 0.66 (0.32), 72mth 0.61 (0.37) ES 0.28), ns. Doering 12 (post) 231 patients Primary outcomes: TFP = 39% (2010) assessed, 127 Drop-out rate: TFP 39%; CP 67 % excluded, 104 Suicide attempts (CISSB): Signif improvement for TFP (Baseline Mean 18 (SD 35.3), 12 mths 7 (SD Experts = RCT allocated to 13.7) ES 0.31) vs CP (Baseline Me an 12 (SD 23.1), 12 mth 11 (SD 21.2) ES 0.04), p<.01. 67% TFP treatment (TFP n=52, Secondary outcomes: (n=52) vs. CP n=52). Depression (BDI): No signif difference between TFP (Baseline Mean 32.12 (SD 43.69), 12 mths CP (n=52) Discontinued 21.67 (SD 13.25) ES 0.38) vs CP (Baseline Mean 24.31 (SD 10.18), 12 mth 20.02 (SD 13.22) ES 0.42), treatment: TFP n=13, ns. CP n=22. TFP n=43 General psychopathology (BSI, GSI): No signif difference between TFP (Baseline Mean 1.52 (SD and CP n=29 0.71), 12 mth 1.33 (SD 0.80) ES 0.27) vs CP (Baseline Mean 1.41 (SD 0.64), 12 mth 1.27 (SD 0.76) assessed at 12 mths. ES 0.22), ns. NSSI (CISSB): No signif difference between TFP (Baseline Mean 29.33 (SD 59.83), 12 mth 16.94 (SD 30.64) ES 0.31) vs CP (Baseline Mean 34.50 (SD 85.15), 12 mths 22.0 (SD 52.97) ES 0.15), ns Farrell 8 (post), 40 patients screened Outcomes: BPD Psychopathology (BSI): Signif improvement for SFP / No signif difference between SFT +TAU = (2009) 14 (FU) for eligibility, 8 SFP (Baseline Mean 34.75 (SD 7.67), 8 mth 18.81 (SD 9.47), 14 mth 15.75 (SD 9.10) ES 2.08) vs TAU 0% excluded, 32 (Baseline Mean 33.33 (SD 4.77), 8 mth 32.75 (SD 5.90), 14 mth 33.08 (SD 4.56) ES 0.12), p<.001. RFT SFT + randomized (SFT Diagnostic measure of BPD (DIB-R): Signif improvement for SFP (Baseline Mean 8.63 (SD 1.41), 8 TAU = 25% TAU vs. +TAU n=16, TAU mth 3.44 (SD 2.76), 14 mth 3.25 (SD 2.79) ES 3.82) vs TAU (Baseline Mean 9.17 (SD 0.94), 8 mth TAU n=16). Lost to 8.58 (SD 1.51), 14 mth 8.75 (SD 1.29) ES 0.63), p<.001. treatment SFT +TAU General Psychiatric Symptoms (SCL-90): Signif improvement for SFP (Baseline Mean 1/75 (SD n=0, TAU n=4. 0.54), 8 mth 1.26 (SD 0.60) 14 mth 0.96 (SD 0.47) ES 0.91) vs TAU (Baseline Mean 1.84 (SD 0.86), 8 Completed mth 2.01 (SD 0.79), 14 mth 1.93 (SD 0.72) ES -0.20), p<.01. treatment/included Global Assessment of Function Scale (GAF): Signif improvement for SFP (Baseline Mean 48.81 (SD in analyses SFT +TAU 7.04), 8 mth 60.50 (SD 10.17), 14 mth 66.19 (SD 7.51) ES 1.66) vs TAU (Baseline Mean 49.17 (SD n=16, TAU n=12. 5.78), 8 mth 50.08 (SD 5.07); 14 mth 48.25 (SD 5.29) ES 0.16), p<.01. N.B. It was not stated which of the above were primary vs secondary outcomes. 12, 173 patients Primary outcomes: BPD severity (BPDSI-IV): Signif improvement for SFP (Baseline Mean 33.53 (SD SFT = 27% 40 24, referred, 85 1.23), 12 mth 22.18 (SD 1.67), 24 mth 17.77 (SD 1.21), 36 mth 16.24 (SD 1.51)) vs TFP (Baseline 36 (post) excluded, 88 Mean 34.37 (SD 1.23), 12 mth 25.13 (SD 1.76), 24 mth 23.38 (1.79), 36 mth 21.87 (SD 1.71), p<.01. TFP = 51% allocated to Secondary outcomes: treatment (SFT n=45, Quality of Life (EuroQol thermometer): No signif difference between SFP (Baseline Mean 50 TFP n=43). Lost to (3.29), 12 mth 56 (SD 2.52), 24 mth 65 (SD 3.49), 36 mth 64.5 (SD 4.66)) vs TFP (Baseline Mean 55 therapy SFT n=12, (SD 2.72), 12 mth 64 (SD 4.85), 12 mth 69 (SD 4.85), 36 mth 67.5 (SD 2.91)), ns. TFP n=22. Included in Quality of Life (WHOQOL): No signif difference between SFP (Baseline Mean 10.33 (SD 0.19), 12 analyses SFT n=44, mth 11.17 (SD 0.26), 24 mth 11.42 (SD 0.36), 36 mth 11.59 (SD 0.29)) vs TFP (Baseline Mean 10.42 TFP n=42. (0.09), 12 mth 11.17 (SD 0.19), 24 mth 11.23 (SD 0.26), 36 mth 11.09 (SD 0.19)), ns. N.B. Effect sizes unable to be calculated due to spurious SDs Gratz 3 (post) 24 participants Outcomes: ERGT = 8% (2006) recruited. One Self-harm (DSHI): Signif improvement for ERGT (Baseline Mean 18.58 (SD 26.63), post 5.00 (SD participant from each 4.94) ES 0.51) vs TAU (Baseline Mean 20.67 (SD 25.73), post 30.33 (SD 35.08) ES -0.38), p<.05. TAU = 10% Pilot RCT treatment group BPD symptoms (BEST): Signif improvement for ERGT (Baseline Mean 27.67 (SD 12.11), post 25.83 ERGT + dropped out (SD 5.72) ES 0.15) vs TAU (Baseline Mean 37.30 (SD 11.91), post 34.70 (SD 10.81) ES 0.22), p<.01. TAU (dropout rate = 8%). Emotion regulation (DERS): Signif improvement for ERGT (Baseline Mean 127.92 (SD 19.99), post (n=12) vs. Final sample size 79.95 (SD 23.97) ES 2.40) vs TAU (Baseline Mean 119.90 (SD 20.86), post 115.80 (SD 16.74) ES TAU ERGT n=12, TAU 0.20), p<.01. (n=10) n=12. Depression (DASS): Signif improvement for ERGT (Baseline Mean 19.17 (SD 8.29), post 9.00 (SD 6.52) ES 1.23) vs TAU (Baseline Mean baseline 23.60 (SD 12.82), post 23.20 (SD 15.32) ES 0.03), p<.05. Acceptance & Action Questionnaire (AAQ): Signif improvement for ERGT (Baseline Mean 44.33 (SD 8.14) post 31.58 (SD 4.83) ES 1.57) vs TAU (Baseline Mean 45.60 (SD 5.02), post 45.40 (SD 4.20) ES 0.04), p<.01. 41 N.B. It was not stated which of the above were primary vs secondary outcomes. Gratz 3 (post) 91 assessed for Outcomes: ERGT = 16% (2014) 6 (FU) eligibility, 30 NSSI (DSHI): Signif improvement for ERGT (Baseline Mean 68.47 (SD 159.96), 3 mth 16.67 (SD 12 (FU) excluded, 61 39.74) ES 0.32) vs TAU (Baseline Mean 23.47 (SD 37.19), 3 mth 19.26 (SD 24.25), ES 0.11) p<.05. TAU = 10% RCT randomized (ERGT BPD symptoms (BEST): No signif difference between ERGT (Baseline Mean 33.32 (SD 11.32), 3 mth ERGT + n=31, TAU n=30). 27.47 (SD 6.59) ES 0.52) vs TAU (Baseline Mean 38.06 (SD 10.15), 3 mth 35.88 (SD 6.59) ES 0.21) TAU Intent-to-treat ns. (n=31) vs sample and final Depression (DASS–D): Signif improvement for ERGT (Baseline Mean 20.09 (SD 12.61), 3 mth 13.04 TAU sample size ERGT (SD 5.63) ES 0.56) vs TAU (Baseline Mean 21.39 (SD 14.92), 3 mth 21.30 (SD 5.63) ES 0.01), p<.05. (n=30) n=31, TAU =30. Quality of Life Inventory (QOLI): Signif improvement for ERGT (Baseline Mean -0.88 (SD 2.10), 3 mth 0.31 (SD 1.62) ES 1.19) vs TAU (Baseline Mean -0.57 (SD 2.20), 3 mth -0.50 (SD 1.62) ES 0.07), p<.05 N.B. It was not stated which of the above were primary vs secondary outcomes. Improvement reported to continue to 12 mth FU for NSSI, BPD symptoms and quality of life. Jorg- 6, 111 patients Outcomes: MBT = 33% ensen 12, allocated to General Psychiatric Symptoms (SCL-90-R): No signif difference between MBT (Baseline Mean 1.7 (2012) 18, treatment (MBT (SD 0.6), 6mth 1.6 (SD 0.7), 12mth 1.3 (SD 0.6), 18mth 1.3 (SD 0.7), 24mth 1.2 (SD 0.8) ES 0.83) vs TAU = 30% 24 (post) n=74, TAU n=37). SGT (Baseline Mean 2.0 (SD 0.6), 6mth 1.5 (SD 0.8), 12mth 1,5 (SD 1.0), 18mth 1.4 (SD 0.9), 24mth RCT Drop-out prior to 1.4 (SD 0.8) ES 1), ns. commencing Com- treatment MBT n=16, Depression (BDI-II): No signif difference between MBT (Baseline Mean 31.5 (SD 10.7), 6mth 27.8 bined SGT n=10. 85 entered (SD 11.7), 12mth 24.6 (SD 12.2), 18mth 22.9 (SD 12.7), 24mth 18.8 (SD 11.5) ES 1.19) vs SGT MBT SGT treatment (MBT (Baseline Mean 37.5 (SD 10.6), 6mth 29.8 *SD 16.0), 12mth 25.9 (SD 16.0), 18mth 22.4 (SD 18.0), (n=27) n=58, SGT n=27) and 24mth 22.8 (SD 13.7) ES 1.39), ns. were included in BPD Symptoms (SCID-BPD): No signif difference between MBT (Baseline Mean 6.7 (SD 1.2), 24mth analyses. 2.8 (SD 2.5) ES 3.25) vs SGT (Baseline Mean 6.9 (SD 1.3), 24mth 3.6 (SD 2.1) ES 2.54), ns. Anxiety (BAI): No signif difference between MBT (Baseline Mean 18.6 (SD 9.0), 6mth 17.8 (SD 11.4), 12mth 14.8 (SD 9.3), 18mth 14.7 (SD 10.4), 24mth 13.5 (SD 10.7) ES 0.57) vs SGT (Baseline Mean 23.7 (SD 11.2), 6mth 18.0 (SD 10.7), 12mth 18.7 (SD 14.9), 18mth 17.7 (SD 17.0), 24mth 15.6 (SD 10.1) ES 0.72), ns. Social Adjustment Scale (SAS): No signif difference between MBT (Baseline Mean 2.6 (SD 0.4), 24mth 2.2 (SD 0.5) 0.4) vs SGT (Baseline Mean 2.8 (SD 0.6), 24mth 2.1 (SD 0.6) ES 1.17), ns. Interpersonal Problems (IIP): No signif difference between MBT (Baseline Mean 1.7 (SD 0.6), 24mth 1.2 (SD 0.6) ES 0.83) vs SGT (Baseline Mean 1.9 (SD 0.6), 24mth 1.3 (SD 0.8) ES 1), ns. 42 Global Assessment of Functioning, Social Function (GAF-F): Signif improvement for MBT (Baseline Mean 46.6 (SD 7.5), 12mth 50.5 (SD 8.1), 24mth 56.7 (SD 11.7) ES 1.35) vs SGT (Baseline Mean 44.6 (SD 8.2), 12mth 47.3 (SD 9.6), 24mth 51.3 (SD 11.7) ES 0.82), p<.001. Global Assessment of Functioning, Symptoms (GAF-S): Signif improvement for MBT (Baseline Mean 43.0 (SD 2.8), 12mth 48.5 (SD 6.1), 24mth 59.5 (SD 12.6) ES 5.89) vs SGT (Baseline Mean 43.1 (SD 3.), 12mth 48.0 (SD 7.7), 24mth 54.0 (SD 10.5) ES 2.87), p<.001. N.B. It was not stated which of the above were primary vs secondary outcomes. Koons 3 56 patients referred, Parasuicidal behavior (PHI): No signif difference between DBT (Baseline Mean 5.1 (SD 13.2), 3 mth DBT = 23% (2001) 6 (post) 23 excluded, 33 1.6 (SD 3.7), 6 mth 0.40 (SD 1.3), ES 0.35) vs TAU (Baseline Mean 0.7 (SD 1.3), 3 mth 1.1 (SD 2.3), assessed for 6mth 1.0 (SD 2.2), ES -0.28), ns TAU = 17% RCT eligibility, 5 excluded, Suicidal ideation (BSSI): Signif improvement for DBT (Baseline Mean 36.2 (SD 13.5), 3 mth 34.9 (SD DBT 28 allocated to 13.5), 6mth 26.2 (SD 8.0), ES 0.74) vs TAU (Baseline Mean 44.6 (SD 11.4), 3 mth 41.9 (SD 13.3), (n=10) vs treatment. Eight 6mth 41.5 (SD 14.3), ES 0.27), p<.05 TAU participants did not Hopelessness (BHS): Signif improvement for DBT (Baseline Mean 11.9 (SD 6.7), 3 mth 0.4 (SD 7.5), (n=10) commence/dropped 6 mth 5.1 (SD 5.3), ES 1.01) vs TAU (Baseline Mean 13.6 (SD 6.8), 3 mth 12.0 (SD 7.8), 6mth 14.2 (D out of treatment. 7.3), ES -0.09), p<.01 Final sample DBT Depression (BDI): Signif improvement for DBT (Baseline Mean 22.8 (SD 11.1), 3 mth 21.3 (SD 13.4), n=10, TAU n=10. 6mth 13.4 (SD 7.5), ES 0.85) vs TAU (Baseline Mean 34.7 (SD 14.6), 3 mth 27.0 (SD 14.6), 6mth 29.3 (SD 17.7), ES 0.37), p<.05 Depression (HAM-D): No signif difference between DBT (Baseline Mean 29.7 (SD 13.7), 3 mth 24.7 (SD 10.1), 6mth 17.1 (SD 5.7) ES 0.92) vs TAU (Baseline Mean 32.6 (SD 9.7), 3 mth 31.1 (SD 11.3), 6mth 24.3 (SD 7.8) ES 0.86), ns Anxiety (HARS): No signif difference between DBT (Baseline Mean 18.4 (SD 7.3), 3 mth 18.1 (SD 8.4), 6mth 19.1 (SD 7.5), ES -0.10) vs TAU (Baseline Mean 27.7 (SD 9.3), 3 mth 25.8 (SD 10.7), 6mth 32.2 (SD 12.4), ES -0.48) ns Anger-In (Spielberger AES): No signif difference between DBT (Baseline Mean 22.9 (SD 5.7), 3 mth 19.3 (SD 5.4), 6mth 17.3 (SD 4.0), ES 0.98) vs TAU (Baseline Mean 20.5 (SD 4.7), 3 mth 18.2 (SD 5.4), 6mth 19.2 (SD 6.2), ES 0.28), ns Anger-Out (Spielberger AES): Signif improvement for DBT (Baseline Mean 18.2 (SD 5.7), 3 mth 17.3 (SD 4.8), 6mth 14.5 (SD 3.9), ES 0.65) vs TAU (Baseline Mean 17.2 (SD 5.8), 3 mth 14.6 (SD 3.1), 6mth 17.9 (SD 6.1), ES -0.12), p<.01 Dissociation (DES): No signif difference between DBT (Baseline Mean 22.3 (SD 15.2), 3 mth 20.0 (SD 16.2), 6mth 13.2 (SD 12.0), ES 0.60) vs TAU (Baseline Mean 41.0 (SD 22.4), 3 mth 29.5 (SD 43 22.5), 6mth 30.6 (SD 23.3), ES 0.46), ns BPD Criteria (SCID): No signif difference between DBT (Baseline Mean 6.8 (SD 1.1), 6mth 3.6 (SD 1.6), ES 2.83) vs TAU (Baseline Mean 6.7 (SD 0.8), 6mth 4.2 (SD 2.3), ES 1.13), ns. Health care utilization (THI): Percentage of inpatient admissions pre-treatment for DBT group 30%, post-treatment reduced to 10%; TAU pre-treatment 20%, post-treatment 10%. Linehan 4, 8 63 participants were Outcomes: DBT = 17% (1991) 12 (post) assessed. 10 dropped Suicide attempt and NSSI (PHI): Signif improvement for DBT (Baseline Mean not provided, 4 mth out during pre- 3.50 (SD 7.88), 8 mth 2.82 (SD 8.13), 12 mth 0.55 (SD 0.94), ES could not be calculated without TAU = 58% RCT treatment baseline mean) vs TAU (Baseline Mean not provided, 4 mth 15.91 (SD 25.02), 8 mth 8.73 (SD DBT assessments (DBT 25.48), 12 mth 9.33 (SD 26.95), ES could not be calculated without baseline mean), p=<.05. (n=22) vs. n=5, TAU n=5), 7 Maintenance in therapy: DBT patients significantly more likely to start individual therapy than TAU TAU dropped following patients (100% compared to 73%, z = 2.75, p <.01). (n=22) pre-treatment Psychiatric inpatient treatment: TAU patients spent significantly more days in psychiatric inpatient assessment (DBT treatment than those receiving DBT (z = 1.70, p <.05). N.B. Mean and SDs not provided. n=3, TAU n=4), 2 DBT Depression (BDI), Hopelessness (BHS), Reasons for Living Inventory (RLI): All other outcomes participants within 4 showed no signif difference between DBT and TAU (Means and SDs not provided). sessions. Final N.B. It was not stated which of the above were primary vs secondary outcomes. sample DBT n=22, TAU n=22. Linehan 4, 6 participants Primary outcomes: DBT = 36% (1999) 8, dropped out Days abstinent from Substance Use (Time-line follow back assessment): Signif improvement for 12 (post), following pre- DBT (Baseline Mean 0.29 (SD 0.26), 4 mth 0.82 (SD 0.19), 8 mth 0.79 (SD 0.28), 12mth 0.89 (SD TAU = 73% RCT 16 (FU) treatment 0.24), 16mth 0.94 (SD 0.17) ES 1.04) vs TAU (Baseline Mean 0.32 (SD 0.29), 4mth 0.46 (SD 0.37), DBT & assessment (DBT 8mth 0.56 (SD 0.27), 12mth 0.62 (SD 0.39), 16mth 0.60 (SD 0.36) ES 0.46), p=<.05. replace- n=1, TAU n=5). 4 in Secondary outcomes: ment DBT did not provide Suicide attempt and NSSI (PHI): No signif difference between DBT and TAU between Baseline and meds any data following 12 mths (Means and SDs not reported). Signif improvements for DBT were reported at follow-up. (n=12) vs. pre-treatment Social Adjustment (GSA), Functioning (GAF), Anger (STAI) All other outcomes showed no signif TAU (including one who difference between DBT and TAU between Baseline and 12 mth (Means and SDs not provided). (n=16) died). Signif improvements for DBT were reported at follow-up. Linehan 4, 64 patients assessed Outcomes: DBT = 27% (2002) 8, for eligibility, 40 Days abstinent from Substance Use (Time-line follow back assessment): Signif improvement for 12 (post), excluded, 24 DBT vs TAU in reducing opiate use (Means and SDs not reported), p<.05 Validation = 44 RCT DBT 16 (FU) randomized (DBT Suicide attempt and NSSI (PHI), Social Adjustment (GSA), Functioning (GAF), Global Symptoms 0% (n=11) vs. n=12, CVT + 12S (BSI): All other outcomes showed no signif difference between DBT and TAU between Baseline CVT + 12S n=12). After and 12 mth (Means and SDs not provided). (n=12) assignment, one DBT N.B. It was not stated which of the above were primary vs secondary outcomes. excluded. Final sample DBT n=11, CVT + 12S n=12. Linehan 0, 4, 8, 186 patients Outcomes: DBT = 19% (2006) 12 (post), assessed for Suicidal and NSSI (SASII): Signif improvement for DBT vs TAU in reduction of suicide attempts 16 (FU), eligibility, 75 (23.1% vs 46%; Means and SDs not reported), p<.05. TBE = 43 % RCT 20 (FU) excluded, 111 Suicide ideation (SASII): No signif diff between DBT (Baseline Mean 51.7 (SD 20.3), 12mth 29.8 (SD DBT 24 (FU) randomized (DBT 24.5), 24mth 24.1 (SD 19.8) ES 1.08) vs CTBE (Baseline Mean 59.9 (SD 21.6), 12mth 32.8 (SD 26.3), (n=52) vs n=60, CBTE n=51). 24mth 31.92 (SD 26.8) ES 1.25), ns. CTBE Lost to Reasons for Living (RLI): No signif difference between DBT (Baseline Mean 2.8 (SD 0.7), 12mth 3.2 group FU/discontinued (SD 1.2), 24mth 3.3 (SD 0.9) ES 0.86) vs CTBE (Baseline Mean 2.7 (SD 0.9), 12mth 3.1 (SD 0.8), (n=49) treatment DBT n=16, 24mth 3.1 (SD 0.8) ES 0.44), ns. CBTE n=35. Included Depression (HAM-D): No signif difference between DBT (Baseline Mean 20.2 (SD 5.9), 12mth 14.0 in analyses DBT n=52, (SD 7.3), 24mth 12.6 (SD 6.8) ES 1.05) vs CTBE (Baseline Mean 21.7 (SD 7.3), 12mth 17.0 (SD 8.2), CBTE n=49. 24mth 14.4 (SD 9.1) ES 0.64), ns. N.B. It was not stated which of the above were primary vs secondary outcomes. McMain 0, 4, 8, 271 assessed for Primary outcomes: Number of Suicidal and NSSI (SASII): No signif difference between DBT DBT=39% (2009) 12 (post), eligibility, 91 (Baseline Mean 20.94 (SD 33.28), 4 mth 10.60 (SD 20.96), 8 mth 8.94 (SD 19.07), 12 mth 4.29 (SD 18 (FU), excluded. 180 9.32) ES 0.50) vs GPM (Baseline Mean 32.19 (SD 81.94), 4 mth 14.02 (SD 43.87), 8 mth 11.44 GPM = 38% RCT 24 (FU), randomized (DBT (37.59), 12 mth 12.87 (SD 51.45). ES 0.24), ns. DBT 30 (FU) n=90, GPM n=90). Severity of Suicidal and NSSI (SASII): No signif difference between DBT (Baseline Mean 4.26 (SD (n=90) vs 36 (FU) Lost at 2.87), 4 mth 2.44 (SD 2.42), 8 mth 1.73 (SD 2.43), 12 mth 1.69 (SD 2.37) ES 0.90) vs GPM (Baseline GPM FU/discontinued DBT Mean 3.67 (SD 2.64), 4 mth 2.24 (SD 2.53), 8 mth 1.59 (SD 2.40), 12 mth 1.32 (SD 2.03) ES 0.89), (n=90) n=42, GPM n=39. ns. Final ITT sample (DBT Secondary outcomes: Psychiatric Hospital Days (THI): No signif difference between DBT (Baseline n=90, GPM n=90). Mean 10.52 (SD 24.42), 4 mth 2.32 (SD 11.92), 8 mth 1.91 (SD 8.57), 12 mth 3.73 (SD 14.90) ES 0.29) vs GPM (Baseline Mean 8.70 (SD 24.91), 4 mth 1.90 (SD 9.32), 8 mth 2.09 (SD 8.26), 12 mth 2.23 (SD 6.55) ES 0.26), ns. General Psychiatric Symptoms (SCL-90R): No signif difference between DBT (Baseline Mean 1.91 (SD 0.77), 4 mth 1.68 (SD 0.79), 8 mth 1.56 (SD 1.29), 12 mth 1.55 (SD 1.25) ES 0.47) vs GPM 45 (Baseline Mean 1.85 (SD 0.76), 4 mth 1.64 (SD 0.83), 8 mth 1.47 (SD 0.83), 12 mth 1.48 (SD 1.14) ES 0.49), ns. Depression (BDI): Signif improvement for DBT / No signif difference between DBT (Baseline Mean 27.19 (SD 12.46), 4 mth 29.06 (SD 15.01), 8 mth 24.16 (SD 15/34), 12 mth 22.18 (SD 16.14) ES 1.2) vs GPM (Baseline Mean 35.40 (SD 10.60), 4 mth 28.28 (SD 13.98), 8 mth 27.55 (SD 15.53), 12 mth 24.83 (SD 14.83) ES 1.0), ns. Quality of Life (EQ-5D): Signif improvement for DBT / No signif difference between DBT (Baseline Mean 57.69 (SD 21.55), 4 mth 60.00 (SD 19.51), 8 mth 63.91 (SD 19.19), 12 mth 63.84 (SD 20.47) ES 0.29) vs GPM (Baseline Mean 55.29 (SD 19.41), 4 mth 59.86 (SD 21.33), 8 mth 59.10 (SD 22.24), 12 mth 59.14 (SD 22.03) ES 0.21), ns. Benefits for participants in each treatment condition were maintained at 2-year follow-up. Verheul (in 92 patients referred, Outcomes: Attrition rates DBT 37%; TAU 77%. DBT = 37% (2003) weeks): 28 excluded, 64 Secondary outcomes: Suicide Attempts (LPC): No signif difference between DBT (Means and SDs 11, randomized (DBT not reported) and TAU. (Attempted suicide: DBT (7%) vs TAU (26%)). ns. TAU = 77% RCT DBT 22, n=31, TAU n=33). NSSI (LPC): Signif improvement for DBT vs TAU (DBT patients displayed reduction in self-harm (n=27) vs 33, Drop-out during pre- behaviours; those in TAU deteriorated; Means and SDs not reported), p<.01. TAU 44, treatment phase DBT Self-damaging impulsive behavior (BPDSI): Signif improvement for DBT vs TAU (Means and SDs not (n=31) 52 (post) n=4, TAU n=2. reported), p<0.05). Completed post (DBT N.B. It was not stated which of the above were primary vs secondary outcomes. n=24, TAU n=23). Effect sizes are shown for within treatment. The significance level is as reported in the paper, and shown for time by group interaction. Effect sizes are most straightforwardly interpreted as one of three categories: small (0.2), moderate (0.5), and large (0.8). It is noted that effect sizes are not ideal for non-normally distributed data, however, they are acceptable and were used in this chapter in order to have a standardised number with which to compare outcomes across studies. Effect sizes were calculated by calculating the difference between the Means at Pre and Post-treatment and dividing by the SD at Pre-treatment. Interventions: CP = Community Psychotherapy; CBT = Cognitive Behaviour Therapy; CT = Cognitive Therapy; DBT = Dialectical Behaviour Therapy; ERGT = Emotion Regulation Group Therapy; MBT = Mentalization-Based Psychotherapy; RST = Rogerian Supportive Therapy; SCM = Structured Clinical Management; SFT = Schema-Focused Therapy; SGT = Supportive Group Treatment; STEPPS = Systems Training for Emotional Predictability and Problem Solving; TAU = Treatment as Usual; TFP= Transference-Focused Psychotherapy;. Outcome measures: AAQ = Acceptance and Action Questionnaire; AES = Spielberger Anger Expression Scale; BAI = Beck Anxiety Inventory; BEST = Borderline Evaluation of Severity over Time; BDI = Beck Depression Inventory I; BDI-II = Beck Depression Inventory II; BPDSI-IV = Borderline Personality Disorder Symptom Index, Fourth Version; BHS = Beck Scale for Hopelessness Hopelessness; BPD-40 = Borderline Personality Disorder Checklist-40; BSI-GSI = Brief Symptom Inventory – Global Severity Index; BSI = Borderline Symptom Inventory; BSSI = Beck Scale for Suicide Ideation; CGI = Clinical Global Impression; CISSB = Cornell Interview for Suicidal and 46 Self-Harming Behaviour – Self Report; DASS = Depression Anxiety Stress Scale; DERS = Difficulties in Emotion Regulation Scale; DIB-R = Diagnostic Interview for Borderline Personality Disorders – Revised; DSHI = Deliberate Self-Harm Inventory; EQ-5D = Euroqol Quality of Life Measure; GAF = Global Assessment of Functioning; GSI = Global Severity Index; HAM-D = Hamilton-Depression Scale; HARS = Hamilton Anxiety Rating Scale; IPP = Inventory of Interpersonal Problems; LPC = Lifetime Parasuicide Count; OAS = Overt Aggression Scale-Modified; PHI = Parasuicide History Interview; QOLI = Quality of Life Inventory; SASII = Suicide Attempt Self Injury Interview; SAS = Social Adjustment Scale; SCID-BPD = Structured Clinical Interview for DSM-IV, Personality Disorders, Borderline Personality Disorder criteria; SCL-90-R = Symptom Checklist 90 Revised; SSHI = Suicide and Self-Harm Inventory; THI = Treatment History Interview; WHOQOL-Brief = World Health Organization Quality of Life Assessment – Brief form; WHOQOL = World Health Organization Quality of Life Assessment. 1.5.1 Cognitive Behavioural Treatment of BPD While the tables above describe the efficacy trials in detail, the following section will highlight the key findings and critique the studies. DBT was the first psychotherapy to be evaluated in a RCT for BPD (Linehan et al., 1991). Participants were randomised to either DBT (n=22) or Treatment as Usual (TAU; n=22). Treatment occurred over 12 months, and assessments were held at six-monthly intervals. Results indicated that compared to those in the TAU condition, those randomised to the DBT condition showed significant reductions in suicide attempts and NSSI and spent significantly less time in psychiatric inpatient treatment. These significant differences in favour of DBT were maintained at one-year follow-up. Two additional RCTs, also with small sample sizes, compared DBT to TAU in a population of women with BPD and substance abuse and found similar results (Linehan et al., 2002; Linehan et al., 1999). These findings have been replicated in the USA in a sample of 20 veterans (Koons et al., 2001) and in the Netherlands by researchers independent of the treatment developer, Marsha Linehan (van den Bosch, Koeter, Stijnen, Verheul, & van den Brink, 2005; Verheul et al., 2003). A limitation in all of these studies was their small sample sizes, which ranged from 22–58. However, despite the small sample sizes these were among the first studies to show that it was possible to treat suicidal behaviour in BPD successfully. There were three efficacy studies where DBT was compared with another active treatment, although some of these comparators had significant limitations (as described below). One of these will be covered below in the section on psychodynamic therapies. The largest study of DBT (n=101) conducted by the treatment developer compared DBT to a control condition of “experts” in the treatment of BPD (Harned et 47 al., 2008; Linehan et al., 2006b). They found that those in the DBT condition were significantly less likely to have made a suicide attempt at the end of treatment, required significantly less hospitalisation, and had a significantly lower drop-out rate from treatment. However, there was no significant difference between groups on suicidal ideation, reasons for living or depression. While this was a robust sample size, it is not clear what treatment those in the “treatment by experts” arm actually received, which raises questions about the nature of the comparison group. As such, there may have been marked variability in the quality of the therapy received for those in the “treatment by experts” arm and this was not assessed. Further, the dose of therapy received by those in the “treatment by experts” arm was markedly less than those in DBT. As such, whilst this trial is described as if DBT was compared with a form of active treatment, the comparison had weaknesses. McMain et al. (2009) completed the largest RCT (n=180) comparing DBT with a treatment called “General Psychiatric Management” (GPM), which is broadly based on psychodynamic therapy with the addition of a medication algorithm. The therapy has now been re-labelled as Good Psychiatric Management (Gunderson & Links, 2014). In the McMain et al. (2009) study, 180 participants with BPD were randomised to either DBT or GPM. The hypothesis was that participants receiving DBT would show greater reductions in the frequency and severity of suicidal and non-suicidal self-injurious behaviours. Results showed that after one year of treatment, both groups significantly improved in the area of suicidal and non-suicidal self-injurious behaviour, health care utilisation, and overall distress, and there was no significant difference between groups. It was not possible to say that the therapies were equivalent in their efficacy as 48 it was not designed as an equivalence trial and hence, was not powered in order to determine whether the therapies were equal. Davidson et al (2006) evaluated Cognitive Behaviour Therapy for BPD in 106 participants. Both CBT plus TAU compared with the control group of TAU alone led to improvements in suicide attempts, inpatient psychiatric hospitalisation or contact with Accident and Emergency. There were no significant differences between the active treatment and the control condition. There was a similar pattern of gradual improvement for participants in either treatment condition and after one year of treatment, significant differences were found on the Brief Symptom Inventory positive distress index, but not for other secondary outcomes. Of note, is that this study had a very high rate of follow-up of participants adding strength to the study findings. Cottraux et al (2009) has also evaluated Cognitive Therapy versus Rogerian Supportive Therapy in a RCT of 65 patients. There was no significant difference by the end of the treatment year. The authors report a number of significant differences at the end of the follow-up period, however, there was marked attrition in retention rates of participants, which puts question marks on the of those findings. Systems Training for Emotional Predictability and Problem Solving (STEPPS) was designed as a supplement to standard therapy and was based on a cognitive method that provides psychoeducation in a group format. STEPPS has been evaluated in an RCT by the treatment developers Donald Black and Nancee Blum, with 124 participants (Blum et al., 2008) and in a replication in the Netherlands (Bos, Bas van Wel, Appelo, & Verbraak, 2010) both of which compared STEPPS with TAU. The Bos et al. (2010) study provided individual therapy consistent with the STEPPS group program and concurrently with the group program. Both studies found that those in the STEPPS 49 group showed significantly greater reduction in both general psychiatric and BPD specific symptomatology than those in the TAU group. To date, this shorter treatment has not been evaluated against another treatment model for BPD, and hence, it is unclear how well the shorter treatment holds up in terms of outcomes. Gratz and Gunderson (2006) developed a 14-week group intervention as an adjunct to current outpatient therapy that targeted emotion dysregulation and emotional avoidance. The group was developed in part in recognition that other treatments are resource intensive and may not easily be implemented in routine clinical settings. The first trial randomised 22 participants to the emotion regulation group and current outpatient therapy (TAU) or to continue with TAU alone. The group consisted of 14 weekly, 1.5-hour sessions. The results showed a significant difference between the groups on emotion dysregulation, experiential avoidance, depression and self-harm behaviour. This first trial was followed up by an RCT with the same design, but with a larger sample size (Gratz, Tull, & Levy, 2014). The results again showed significant effects of the emotion regulation group on self-harm and depression. Schema Focused Therapy (SFT) is an extension of standard CBT and aims to modify maladaptive schemas developed during childhood. To date there has been one published RCT utilising individual therapy (Giesen-Bloo et al., 2006) and one utilising group therapy (Farrell et al., 2009). There is an international multi-site trial of group schema therapy currently underway with a planned sample size of 448 patients (Wetzelaer et al., 2014). A Dutch study compared SFT with Transference Focused Psychotherapy (TFP; Giesen-Bloo et al., 2006) and randomised 44 patients to SFT and 42 patients to TFP. There was a main effect for time, such that participants significantly improved on BPD severity and quality of life across the therapy. However, those in the 50 SFT group showed significantly greater recovery in terms of BPD severity, general psychopathology, and quality of life. While the outcomes from the Dutch study are promising, treatment duration was three years, which means that such treatment is unlikely to be readily accessible to most patients. The SFT group received more supervision than those in the TFP group, lending a bias towards SFT. A further limitation is that all authors were allegiant to SFT, which increases the risk of bias in favour of schema therapy (Luborsky et al., 1999). SFT provided in group therapy format has been evaluated in one RCT by Farrell et al. (2009). They tested the effectiveness of a 30-session group designed to be utilised as an addition to TAU and compared this with TAU alone. A total of 32 patients were randomised, 16 to each condition. All patients randomised to SFT were retained in treatment. Compared to those in TAU alone, those in the SFT group improved significantly on measures of BPD symptoms, the severity of general psychiatric symptoms, and global functioning. Improvements were maintained at six-month follow-up. Although the results of this study are very promising, the authors are the treatment developers and the results need to be replicated with a larger sample size. 1.5.2 Psychodynamic Treatment of BPD Mentalisation Based Therapy (MBT) is the psychodynamic treatment most systematically evaluated in the treatment of BPD. The treatment derives from attachment theory and theory of mind (Fonagy & Target, 2006). The original RCT (Bateman & Fonagy, 1999) was conducted in a partial hospitalisation setting with 38 patients with BPD and found significant improvement across a range of outcomes, including suicide attempts and non-suicidal self-injury, depressive symptoms and general psychiatric symptoms. Given that the treatment occurred in a hospital setting 51 there are a range of other influences on patients that may have impacted the outcomes. Consequently, it is difficult to isolate MBT as the active ingredient in those outcomes. In addition, in many settings, including the current Australian public health setting, treatment with partial hospitalisation would not fit the existing context. Bateman and Fonagy (2009) followed up the original trial with a study in a standard outpatient setting where 134 patients were randomly assigned to either 18 months of MBT or 18 months of structured clinical management. Results showed that patients in both groups significantly improved, but those in the MBT group made significantly greater progress in terms of suicide attempts and hospitalisations. MBT has been evaluated by a research group independent of the treatment developers (Jorgensen et al., 2013) who compared participants receiving MBT with a biweekly supportive group therapy. In a sample size of 58 participants, they found significant changes in both treatment groups for the majority of outcomes. MBT was only superior to the control group in the improvement of Global Assessment of Functioning scores. Transference Focused Psychotherapy (TFP) has been examined in three RCTs. The first was a three-armed trial comparing TFP with DBT and a dynamic supportive treatment (Clarkin et al., 2007). This study was conducted by researchers allegiant to TFP. Their primary outcome domains were suicidality, aggression, and impulsivity and secondary domains were anxiety, depression, and social adjustment. Thirty patients were randomised to one of each of the arms and received one year of treatment. All patients showed improvement; however, the patterns of improvement differed depending on the treatment received. Patients in DBT and TFP showed significant improvement in suicidality, while patients in TFP and supportive therapy showed significant improvements in anger, and only TFP was associated with a significant 52 reduction in impulsivity. There are a number of limitations of the study that are shared across many studies in the area of BPD. These include allegiance of the researchers to one particular model, small sample size, no adherence ratings conducted, and medications not standardised across the three groups. Doering et al. (2010) conducted an RCT of TFP where the comparison treatment group comprised community psychotherapists. They randomised 104 female outpatients and found that significantly fewer participants in the transference-focused arm attempted suicide or dropped out of therapy at the 12-month time-point. However, there were no differences between TFP and the community psychotherapists on secondary outcomes of depression, general psychopathology or NSSI. While low drop out from therapy is important because in order to receive help, people need to stay in therapy; improved outcomes beyond just the important goal of reduction in suicide attempts are needed for participants to have a reduced sense of misery and improved quality of life. In addition, a key limitation is that they only had endpoint data for 68% of the participants and the amount of missing data has implications for the claims that can be made. The other study that explored TFP was the Dutch trial discussed above that compared SFT and TFP (Giesen-Bloo & Arntz, 2007). Another psychodynamic model developed specifically for clients with BPD is Conversational Model (CM; Meares, 2004; Meares & Stevenson, 2000; Meares et al., 1999). CM was the first dynamic therapy for BPD to be empirically evaluated. While a few observational studies have been conducted, to date, CM has not been the focus of an RCT. As such, the existing studies would not be considered as efficacy research, so further information will be included in the section below on CM effectiveness studies. 53 As can be seen in Table 1.3, DBT has the most research support in the treatment of BPD, with six replications following the original trial. With multiple positive outcomes shown across those replications, at this point, DBT can be considered the most reliably well-established of all of the psychotherapies for BPD. Despite this, there is only one study with strong methodological rigour where DBT is compared against an active control group (McMain et al, 2009) and in that trial, there were no differences in outcome between DBT and GPM. 1.6 Effectiveness Studies for DBT and CM There are multiple published effectiveness studies across the evidence-based therapies for BPD. It is beyond the scope of this thesis to cover all of them and discussion will be restricted to coverage of DBT and CM. Given that the current study was conducted in a routine clinical setting of a public sector mental health service with routine therapists, the previous research on these therapies conducted in similar settings will be considered in detail in this section. DBT effectiveness studies were identified by conducting a search in PsycInfo and Medline using the search terms “Dialectical Behaviour Therapy” (US and UK spellings) OR “DBT” AND “restricted to English language publications”. CM effectiveness studies were identified using the same process, using the term “Conversational Model” instead of DBT. No limits were placed on publication year. The list of studies generated was then reviewed for studies conducted with adults with BPD in routine clinical outpatient settings. Further studies were identified by searching the reference lists of studies identified through the literature review and searching for other published studies of treatment of adults with BPD in routine clinical 54 outpatient settings. Information about the study settings, inclusion and exclusion criteria and the participant characteristics is included in Table 1.4. The interventions used in each study including information about the control conditions (where present) are shown in Table 1.5. Information regarding attrition and the outcomes for each study are reported in Table 1.6. The outcomes that were chosen to be included were those that were measured in a number of outcome studies for BPD. As was the case with the efficacy studies, there was marked heterogeneity in the outcomes that were measured and the measures that were utilized. 55 Table 1.4 Participant Characteristics of Effectiveness Studies of Dialectical Behaviour Therapy (DBT) and Conversational Model (CM) for Treatment of Borderline Personality Disorder (BPD) in Adults Authors N Mean in % % Education Inclusion Exclusion Illness factors Study years (SD) Female Ethnicity Welfare criteria criteria setting Marital Status Country Dialectical Behaviour Therapy studies of duration 12 months Comtois et 38 34 (SD not 96% Caucasian Unemployed 82%; Adequate Significant Participants were referred from hospital al (2007) reported) 96 working sporadically English mental inpatient units, community therapists, 9%; working part-time proficiency, retardation involuntary civil commitment proceedings, Community 9% chronic self- or current crisis intervention services, and other local 56 Mental injury and/or substance mental health services. Participants Health Single/never married experience of dependence considered “most difficult to treat” were 52%; divorced 31%; multiple without accepted, that is, individuals with comorbid USA cohabiting with a treatment concurrent diagnoses of bipolar or psychotic disorders. partner 17% failures as treatment. Diagnostic characteristics: BPD 96%; assessed by the primary DSM-IV Axis I disorder of Treatment depression or dysthymia 87%; History schizoaffective disorder 4%; bipolar Interview disorder 4%; schizotypal disorder 4%. (Linehan & Comorbid diagnoses: eating disorder 52%l Heard, 1987). anxiety disorder 65%; substance abuse or dependence 43%; history of self-inflicted injury including > 1 suicide attempt 91%. Feigen- 41 35 (SD not 73% Not Mean number of years 18 to 65 years, Extensive Comorbid current Axis I diagnoses: Major baum et al reported) reported. in education DBT diagnosis of forensic depressive disorder DBT 56%, TAU 50%; (2012) 12.38(SD=1.8), TAU BPD in history with Dysthymia DBT 28%. TAU 25%; Substance 12.4(SD=2.5) accordance with evidence of abuse DBT 40%, TAU 54%; panic disorder Specialist DSM-IV current high with agoraphobia DBT 36% TAU 25%; social service for Unemployed DBT 88%, risk to phobia 0%, TAU 6%; OCD DBT 4%, TAU 6%; Personality TAU 94%; Full-time others, PTSD DBT 32%, TAU 38%; Bulimia DBT 4%. Disorders paid DBT 4%, TAU 0%; diagnosis of TAU 12%. Comorbid Axis II diagnoses: Anti- in National Part-time paid DBT 0%, schizophreni social BT 28%, TAU 25%; Narcissistic DBT Health TAU 6%; Homemaker a or bipolar 4%, TAU 13%; Avoidant DBT 36%, TAU 38%; Service DBT 8%, TAU 0% disorder, Dependent DBT 8%, TAU 0%; Obsessive- primary compulsive DBT 4%, TAU 0%; Paranoid DBT United Single DBT 80%, TAU diagnosis of 40%, TAU 13%; Schizotypal DBT 0%, TAU Kingdom 88%; Married DBT 8%, substance 7%; Schizoid DBT 0%, TAU 7% TAU 0%; Cohabiting abuse, DBT 4%, TAU 0%; severe Separated/divorced cognitive 8%, TAU 12% impairment Hjalmars- 27 20.2 (5.6) 100% Not 19 of the 27 patients Female, meet Psychosis, All participants met criteria for > one son (2008) reported. answered questions criteria for BPD severe eating additional Axis II diagnosis (in accordance regarding civil status in accordance disorder, with DSM-IV). Comorbid Axis II diagnoses: 57 Psychiatric and highest degree of with DSM-IV drug Depressive PD 905; antisocial PD 60%; health care education. Of those: addiction paranoid PD 60%; phobic PD 55%; passive- service 79% completed aggressive PD 50%. compulsory school Sweden 89% unmarried Priebe 80 32.2 (10.8) 88% White Education not 5+ days with Severe Participants were recruited from a range of (2012) & 57.5%; reported. self-harm in the learning primary, secondary and tertiary services. Barnicot Black 15%; Regular employment year prior to difficulties, (2013) Asian 36.3%; treatment, 16+ lack of The average number of self-harm days in 21.3%; Voluntary/protected/ years, diagnosis English the previous 2 months was 13.9(18.4), National Mixed/oth sheltered work 10%; of at least one proficiency average number of suicide attempts in past Health er 6.3% no work 53.8%; personality 12 months was 2.2(6.0). Service Independent disorder accommodation 85%; The average number of comorbid Axis I United sheltered or supported disorders was 8.0(3.1) and average number Kingdom accommodation 7.5%; of Axis II disorders was 3.5(1.6). homeless or 24h supervised 7.5% Marital status not reported. Stiglmayr 70 30.1 (8.1) 92% Not No formal education 16+ years of Lifetime Participants recruited through a range of (2014) reported. 2.1%; 9 year 8.5%; 10 age, diagnosis diagnoses of inpatient and outpatient services. year 59.6%; high- of BPD in schizophreni Routine school 29.8% accordance with a, bipolar I Comorbid diagnoses (current) included: any Mental DSM-IV-TR, disorder, depressive disorder 38.3%; panic disorder Health Retired due to medical acute with agoraphobia 14.9%; social phobia Care reasons 19.1%; suicidality, 23.4%l obsessive compulsive disorder 6.4%; unemployed 44.7%; substance PTSD 36.2%; other anxiety disorder 23.4%; Germany student 17.0%; dependence any anxiety disorder 61.7%; eating disorder working/employed within last 6 36.3%; substance abuse 25.5%; any Axis-I 19.1% months, BMI disorder 91.5%; other personality disorder < 18, IQ < 80, 68.1%. Single 38.3%; married diagnosis of or in a steady ASPD 58 relationship 61.7% Dialectical Behaviour Therapy studies of duration 6 months or less N.B. (only exception was Axelrod & Williams study which were 20 weeks) Axelrod 27 38 (SD not 100% Caucasian Some college Female, met Active A public-funded clinic delivered treatment, (2011) reported) 92%, education 31%; high DSM-IV criteria psychosis, providing services only to low Hispanic school education 31% for BPD and schizophren- socioeconomic status individuals with no Substance 8% substance ia, bipolar insurance or with public insurance only. use clinic Never married 62%; dependence, disorder, divorced 38% consecutive actively Average number of comorbid Axis I USA admissions to a suicidal, lack diagnoses was 2.4 (81% depressive primary of English disorders, 69% anxiety disorders, 50% substance use proficiency PTSD, 6% bipolar disorder). clinic Ben-Porath 26 35.48 (SD 96% Caucasian Unemployed 69.23%; BPD & co- Nil reported Participants recruited from community (2004) 10.19) 100% receiving government morbid severe mental health setting exclusive for assistance 88.46% Axis I mental treatment of severe mental illnesses. Community illness. Mental Comorbid Axis I diagnoses: bipolar disorder Health 30%; major depression 44%; schizoaffective disorder 22%; schizophrenia 4% USA Blenner- 9 29.4 (SD 100% Not Not reported. 18 to 65 years, Organic brain Participants referred from community hassett not reported. attending St disorder, mental health teams. (2009) reported) Ita’s outpatient substance services, dependence, Comorbid lifetime Axis I diagnoses: 67% Community diagnosis of significant major depressive disorder, 22% bulimia Mental BPD in physical nervosa, 11% schizophrenia, 56% Health accordance with illness substance abuse. DSM III-R Ireland Brassing- 10 34.3 (SD 100% New Not reported. Diagnosis of Not reported. Comorbid diagnoses identified by ton (2006) not Zealand/ BPD in administration of the Millon Clinical reported) Pakeha accordance with Multiaxial Inventory, 3rd edition (MCMI-III). Community 90%; International Scores >85 indicate severe prominence of a Mental Maori/ Personality syndrome/disorder, scores between 75 & 59 Health European Disorder 85 indicate moderate severity, scores <75 10% Examination indicate not clinically significant presence New (IPDE) of syndrome/disorder. Pre-treatment Zealand means for comorbidity on MCMI-III scales were as follows: anxiety 92.70; somatoform 76.20; bipolar: manic 69.60; dysthymia 91.70; alcohol dependence 73.20; drug dependence 61.60; PTSD 83.20; thought disorder 72.50; major depression 97.30; delusional disorder 68.20. Severe personality pathology included schizotypal 80.30; borderline 90.50; paranoid 76.50. Carter 73 24.5 100% Not Post-school Female, 18 to Disabling Participants were referred by Hunter (2010) (SD 6.10) reported. qualification 38.6%; 65 years, met organic Mental Health Services, general completed secondary criteria for BPD condition, practitioners, and treating psychiatrists. Specialist education 22.9%; did in accordance diagnosis of service for not complete with DSM-IV, schizophrenia Comorbid diagnoses included: substance BPD within secondary education history of , bipolar use disorder 68.6%; any affective disorder public 34.3%; still at school multiple affective 73.0%; any anxiety disorder 88.6%. Mean 4.3% episodes of self- disorder, mental harm (>3 within psychotic number of lifetime self-harm episodes was health Employed 18.6%; the preceding depression, 96.3 (SD-109.70). unemployed 5.7%; not 12 months) florid Australia in labour force 75.7% antisocial behaviour, Married 25.7%; development divorced 18.6%; single al disability 55.7% Pasieczny 90 33.58 (SD 93% Not Senior education Current patient No specific Participants recruited from large Australian (2011)1 10.10) reported. 41.11%; junior of Princess exclusion inner city integrated public mental health education 58.89% Alexandra criteria. service. Public Hospital Participants Mental Employed 21.11% Division of deferred if Comorbid diagnoses: schizophrenia or Health Mental Health, unable to bipolar 10%, depression 76.67%, PTSD 60 service Married/defacto 18+ years, met complete 23.33%, anxiety 50%, substance 24.44% criteria for BPD, assessments dependence 25.56%, substance abuse Australia ongoing due to 51.11% suicidal/NSSI, > imminent risk 1 acute of suicide or psychiatric acute hospitalisation psychosis. or 3 ED presentations in last 6 months, agreement to participate in evaluation Prender- 11 36.35 (SD 100% Not Unemployed 73% 18+ years, met Current >1 comorbid Axis I diagnosis was present in gast (2007) 7.42) reported. criteria for BPD psychotic 64% of participants, 27% had two comorbid single 34%; in accordance episode, Axis I diagnoses. Most frequent comorbid married/defacto 45% with DSM-IV diagnoses were dysthymia 45%; major 1 n.b. The self-report measures were not introduced until after 45 patients had been recruited, hence, the outcomes reported based on self-report measures are for n=45. Public substance depressive disorder 18%; PTSD 9%; bipolar mental dependence affective disorder 9% health & women’s health service Australia Williams 31 Mean for 83% Not Not reported. Diagnosis of Current, During 6 months prior to treatment, 21% of (2010) overall reported. BPD in severe and participants were in psychiatric inpatient sample of accordance with uncontrolled care for 1+ days, 66% presented to Public group DSM-IV-TR, psychotic emergency departments, 37% made 1+calls mental program commitment to illnesses, to ACIS. health completers attend group severe 61 service across DBT program or substance Comorbidity statistics not reported. and TAU TAU, residence abuse, Australia (N=68); within service significant 35.59 catchment area aggressive or (SD=10.02) antisocial traits Conversational Model studies Stevenson 30 29.4 (SD 63% Not Not completed high Diagnosis of Nil specified. Participants referred from a range of (1992) 7.9) reported. school 10%; completed BPD. inpatient and outpatient community college studies 20% services, as well as self-referred. Public teaching Received long-term Comorbidity statistics not reported. hospital institutional or foster 26.67%; receiving Australia government financial assistance 73.3%; professional employment 10%; skilled employment 10%; nonskilled employment 6.67% Married 30% Korner 29 27.9 (SD 59% Not For overall group Diagnosis of Nil specified. Participants referred from the Westmead (2006) 5.9) reported. (including TAU BPD. Personality Disorder Research and participants): Treatment Program. Public unemployed/receiving teaching government benefits Comorbidity statistics not reported. hospital 78%; employed 21.67%; Australia Left school before Year 10 81.67%; tertiary 62 study 18.33% Received long-term institutional or foster care 36.67% Table 1.5 Intervention Descriptions of Effectiveness Studies of Dialectical Behaviour Therapy (DBT) and Conversational Model (CM) for Treatment of Borderline Personality Disorder (BPD) in Adults Authors Intervention Summary Supervision Session frequency / length Summary of control Attendance Study Consultation Duration sessions Design Attendance Delivered by Dialectical Behaviour Therapy studies of duration 12 months Comtois et DBT program delivered Adherence monitored through Session frequency/length: DBT Not applicable. Not reported. al (2007) in accordance with weekly individual supervision involved twice-weekly skills groups (90 manualized treatment provided by a senior DBT mins duration each), individual weekly Pre-post developed by Linehan trainer, in addition to periodic sessions (60 mins duration), phone (1993) focusing on training conducted by DBT consultation as needed, case- 63 balancing acceptance therapists themselves. management as needed, medication with change strategies management (30 mins every 1 to 3 Slight adaptations to months). Duration: 1 year. Delivered standard DBT included by: 1 psychiatrist, 3 psychologists, 2 DBT oriented case masters-level clinicians, 1 bachelor’s management and level clinician (average DBT experience medication was 7.7 years). management. Feigenbaum Manualised DBT as per Adherence not formally Session frequency/length: DBT TAU involved standard Not reported. et al (2012) the Linehan (1993) assessed, however therapists involved weekly individual sessions (60 care provided by the treatment manual was adherence to DBT model mins duration), and weekly group skills referring agency, RCT of implemented. Treatment monitored through weekly case training (150 mins duration), in including: outpatient DBT vs. TAU focused on reducing self- discussions, verbal reporting of addition to phone consultation as psychiatric review, injurious behaviours and session content, and reviewing required. Duration: 1 year. Delivered case management, working towards a life one another’s session audio by: 3 clinical psychologists, 2 senior psychoanalytic worth living through tapes. community psychiatric nurses, 1 psychotherapy, integration of both consultant psychiatrist cognitive behavioural acceptance and change therapy, supportive strategies. Session frequency/length: TAU structured counselling, involved a range of individual services, inpatient admissions, with length and frequency varying drug and alcohol between service providers. Duration: 1 treatment, and crisis year. Delivered by: unspecified (mental management. health professionals). Hjalmarsson Manualised DBT as per Adherence was monitored Session frequency/length: DBT Not applicable. Not reported. (2008) the Linehan (1993) through ensuring session involved weekly individual sessions (6o treatment manual was content and hierarchy matched mins duration), weekly group skills Pre-post implemented. Treatment the treatment manual. training (180 mins duration). focused on reducing self- Therapists received 3 hours of Duration:1 year. injurious behaviours and weekly group supervision. All Delivered by: 2 physicians, 3 working towards a life sessions were videotaped to psychologists, 8 registered nurses, 8 worth living through ensure treatment adherence. mental health assistants, 1 integration of both Adherence also assessed occupational therapist acceptance and change through the DBT adherence- 64 strategies. scale during individual supervision sessions. Priebe DBT delivered in Adherence was monitored Session frequency/length: DBT TAU involved Average length (2012) & accordance with through audio recordings of involved weekly individual sessions (60 engagement in any of stay in Barnicot Linehan’s (1993) 10% treatment sessions mins duration), weekly group skills treatment other than treatment for (2013) treatment manual. assessed using a 63-item training (120 mins duration), phone DBT within their local DBT group was Treatment focused on adherence scale (scores range consultation as required. Duration: 1 community. 5.7 months (SD Pragmatic achieving positive from 0 to 5; 4.0 indicative of year. Delivered by: DBT therapists = 3.2). RCT of DBT change whilst balancing good adherence). Five group (details not reported). vs. TAU this with acceptance and sessions also rated for supportive therapy adherence using the same Session frequency/length: TAU techniques. scale. involved a range of a range of individual services, with length and frequency varying between service providers. Duration: 1 year. Delivered by: psychotherapists, psychiatrists, community mental health teams, counsellors, general practitioners, user-run support groups Stiglmayr DBT was delivered in Adherence monitored using the Session frequency/length: DBT Not applicable. The average (2014) accordance with Adherence Coding Scale (scores involved weekly individual sessions (50 number of Linehan’s (1993) range from 0 to 5; >4 indicative min duration), weekly skills group therapy Pre-post treatment manual. of good adherence). Two training (120 min duration), phone sessions was Treatment focused on randomly selected videotapes consultation as required. Duration:1 38.5 (SD=9.2). achieving positive were rated for adherence year. Delivered by: 20 For skills change whilst balancing during first 3 months of psychotherapists with DBT experience training groups, this with acceptance and treatment; and 1 videotape in for an average as 5.9 years (SD=3.4). 89.4% of supportive therapy the proceeding 3-month period participants techniques. (total of 5 video-tapes per attended an therapy reviewed). Mean rating average of 18.9 for adherence was 4.17 sessions (SD=0.12). (SD=10.0). Dialectical Behaviour Therapy studies of duration 6 months or less N.B. (only exception was Axelrod & Williams study which were 20 weeks) 65 Axelrod DBT was adapted for the Adherence was monitored Session frequency/length: DBT Not applicable. Average (2011) needs of the substance through a DBT trainer attending involved weekly individual sessions (60 number of use facility providing consultation meetings between min duration), weekly skills group individual Pre-post therapy. Aims of therapists, and supervision of training (90 min duration), phone therapy treatment involved the 1 doctoral level and 3 consultation as required. Duration: 20 sessions reducing substance use masters level clinicians weeks. Delivered by: a range of attended was (considered a quality-of- providing treatment. clinicians (including 3 masters level 14.9, and life-interfering behavior), clinicians and 1 doctoral level average number in addition to reducing clinician). Training provided by DBT of group life-threatening and expert. sessions treatment-interfering attended was behaviours. 14.0. Ben-Porath DBT delivered in Adherence was not formally Session frequency/length: DBT Not applicable. Attendance for (2004) accordance with assessed. Therapists attended involved weekly individual sessions (60 individual Linehan’s (1993) weekly peer supervision, part of min duration), weekly group skills therapy was Pre-post treatment manual. which involved discussion of training (90 min duration), phone 89%; 81% for (n=26) Treatment focused on fidelity to DBT framework. consultation as required, psychiatric group skills achieving positive services, and case management. training change whilst balancing Duration: 1 year. Delivered by: 2 sessions. this with acceptance and psychiatrists, 2 psychologists, 1 supportive therapy licensed professional clinical techniques. counselors, 3 master-level counselors Blennerhass DBT was delivered in Adherence was not formally Session frequency/length: DBT Not applicable. Not reported. ett (2009) accordance with assessed. Prior to each skills involved weekly individual sessions (60 Linehan’s (1993) training group, DBT therapists min duration), weekly group skills Pre-post treatment manual. delivering treatment would training (120 min duration), phone (n=9) Treatment focused on meet for 1 hour to discuss consultation as required. Duration: 6 achieving positive group preparation. months. change whilst balancing Delivered by: 3 experienced DBT this with acceptance and therapists. supportive therapy techniques. Brassington DBT was delivered in Adherence was not formally Session frequency/length: DBT Not applicable. Not reported. (2006) accordance with assessed. Weekly group involved weekly individual sessions 66 Linehan’s (1993) consultation included (60-90 min duration), group skills Pre-post treatment manual. treatment review and group training (150 min duration), phone (n=10) Treatment focused on supervision. consultation as required. Duration: 6 achieving positive months. change whilst balancing Delivered by: DBT therapists (details this with acceptance and not reported). supportive therapy techniques. Carter DBT was delivered in Adherence was not formally Session frequency/length: DBT TAU involved a 6- Not reported. (2010) accordance with reported on. Therapists involved individual sessions, group- month waitlist for DBT Linehan’s (1993) involved participated in an based skills training, and phone whilst continuing with RCT of DBT treatment manual. introductory programme in consultation as required. Length of current treatment. vs. WL + TAU Treatment focused on addition to group DBT sessions not reported. Duration: 6 (n=73) achieving positive consultation sessions. months. Delivered by: psychiatrists, change whilst balancing social workers, clinical psychologists, this with acceptance and psychologists, occupational therapists supportive therapy or nurses. techniques. One distinction from standardized DBT involved the use of a group roster of DBT individual therapists between 8:30am and 10pm for phone coaching. Pasieczny DBT was delivered in Adherence was assessed by Session frequency/length: DBT TAU involved clinical Average (2011)2 accordance with analysis of recorded individual involved weekly individual sessions (60 case management number of face- Linehan’s (1993) therapy sessions using a DBT min duration), weekly group skills from case managers to-face contacts DBT vs. WL / treatment manual. adherence checklist (an training (120 min duration), and phone across a range of with individual TAU, but not Treatment aimed to adapted version of therapy consultation as required. Duration: 6 disciplines (psychiatric therapist/case randomised; reduce life-threatening, checklist (Linehan, 1993) months. Delivered by: 18 DBT nurses, psychologists, manager among based on therapy-interfering, and Cognitive Therapy for therapists (including 3 masters trained social workers, DBT group was 67 group quality-of-life interfering Borderline Personality Disorder. psychologists, 4 generalist occupational 18.85 (SD=3.36), availability. behaviours. Behaviour (mean adherence =10.75) psychologists, 5 social workers, 2 therapists). A range of TAU was 19.18 (n=90) change is complemented consultant psychiatrists, 1 psychiatric services were (SD=6.58). No by increasing skillfulness registrar, 2 occupational therapists, 1 provided, for example significant in mindfulness, emotion clinical nurse). engagement, difference regulation, distress assessment, linking revealed. tolerance, and with services, crisis interpersonal management, effectiveness. counselling and skills training. Prendergast DBT delivered in Adherence was not formally Session frequency/length: DBT Not applicable. Not reported. (2007) accordance with assessed. Therapists attended involved weekly individual sessions Linehan’s (1993) either weekly supervision for 90 (60-90 min duration), weekly group Pre-post treatment manual. mins, or bimonthly supervision sessions (150 min duration), phone (n=16) Adaptations to DBT as for 120 mins. consultation as required. Duration: 6 per treatment manual months. 2 n.b. The self-report measures were not introduced until after 45 patients had been recruited, hence, the outcomes reported based on self-report measures are for n=45. included: female-only Delivered by: 12 female therapists (5 closed group, telephone psychologists, 2 counsellors, 2 consultation support occupational therapists, 1 clinical from hospital staff in psychiatric nurse, 1 social worker). addition to individual therapists, and a shorter course of treatment. Williams DBT was delivered in Adherence was not formally Session frequency/length: DBT TAU involved 61% of (2010) accordance with assessed. DBT therapists included weekly individual sessions, continued treatment participants Linehan’s (1993) attended regular group and weekly group sessions (120 min with one’s individual completed Pre-post treatment manual. consultation. duration). Phone consultation not therapist. individual DBT, (n=31) Treatment focused core provided by individual therapists. compared DBT modules: emotional Duration: 20 weeks. Delivered by: DBT- to43% of regulation, interpersonal trained clinical psychologists. participants 68 effectiveness, receiving TAU. mindfulness and distress tolerance. Adaptations from the treatment manual included group psychoeducation regarding BPD in session 1, group celebration during the final session, and developing individual crisis management plans at end of treatment. Conversational Model studies Stevenson Treatment based on Adherence was achieved Session frequency/length: not Not reported. Not reported. (1992) psychology of the self, through providing weekly reported. Duration: 12 months. informed by Kohut seminars focusing on treatment Delivered by: trainee therapists closely Pre-post (1984) and Winnicott framework, audio-recordings of supervised by experienced (n=30) (1971). Treatment all sessions reviewed during psychotherapists. therefore focused on weekly supervision, and peer development and supervision for all supervisors. expression of the self A linguistic analysis of session through 2 main forms of transcripts was also conducted. conversation (1) Maturational, achieved through coupling, amplifying and representing the self (2) Transference within the therapeutic relationship of unconscious traumatic memory. Korner See above – Stevenson Adherence was assessed Session frequency/length: TAU involved Not reported. 69 (2006) (1992). through the use of audio-taped Conversational Model involved twice- continued care as sessions reviewed during weekly individual sessions (50 min usual, and crisis Un- supervision, in addition to duration). Participants also required to support as required. controlled attendance at theoretical maintain contact with their case TAU participants were WL + TAU lectures and seminars. managers should they require crisis offered Conversational (n=29) Therapists attended supervision support. Duration: 12 months. Model following a with therapists experienced in Delivered by: trainee psychotherapists, waitlist period. the Conversational Model. with training in psychiatry or general medicine. CM = The Conversational Model; DBT = Dialectical Behaviour Therapy; RCT = Randomised Controlled Trial / Randomised Clinical Trial; TAU = Treatment as Usual; WL = Waitlist. Table 1.6 Summary of Results of Effectiveness Studies of Dialectical Behaviour Therapy (DBT) and Conversational Model (CM) for Treatment of Borderline Personality Disorder (BPD) in Adults Authors Assess- Attrition from study Outcome measures and results Drop-out Study ment rates from Design points treatment Dialectical Behaviour Therapy studies of duration 12 months Comtois et 0, 12 38 patients enrolled Primary outcomes: Medically treated parasuicides (THI): No signif difference between Baseline DBT = 24% al (2007) (post) in treatment, 9 2.91 (SD 6.25) and 1 year 2.09 (SD 8.30), ES 0.13, ns months dropped-out in first Psychiatric-related emergency room visits (THI): No signif difference between Baseline 7.09 (SD Pre-post 4mths, 4 unable to 12.16) and 1 year 5.04 (SD 13.80), ES 0.17, ns (n=38) be Psychiatric inpatient admissions (THI): Signif improvement between Baseline 5.13 (SD 6.43) and 1 70 contacted/relocated, year 1.43 (SD 2.83), ES 0.58, p<.01 1 not included in Crisis-related inpatient admissions (THI): Signif improvement between Baseline 3.26 (SD 4.86), 1 analyses as year 0.61 (SD 1.50) , ES 0.55, p<.01 transferred to Psychiatric-related inpatient days (THI): Signif improvement between Baseline 39.22 (SD 51.68), 1 advanced program year 10.83 (SD 27.22) , ES 0.55, p<.01 following 6mths. 1 No. crisis systems engaged (THI): Signif improvement between Baseline 2.50 (SD 1.77) 1 year 1.45 patient died prior to (SD 1.90) , ES 0.59, p<.01 completion of first Cost of psychiatric-related inpatient hospitalizations (THI): Signif improvement between Baseline year of treatment. $29,843 (SD $38,693), 1 year $8,140 (SD $20,308) , ES 0.55, p<.01 Feigenbau 0, 6, 12 124 patients Primary outcomes: DBT = 56% m et al (post) referred, 60 eligible General Psychiatric Symptoms (CORE-OM): No signif difference between DBT (Baseline Mean 2.59 TAU = 6% (2012) months to participate, 18 (SD 0.69), 6mth 2.37 (SD 0.83), 12mth 2.34 (SD 0.87) ES 0.36) vs TAU (Baseline Mean 2.40 (SD withdrew before 0.56), 6mth 2.27 (SD 0.84), 12mth 2.19 (SD 0.86) ES 0.38), ns. RCT of randomization, 42 CORE-OM (Risk): No signif difference between DBT (Baseline Mean 1.57 (SD 0.97), 6mth 1.38 (SD DBT vs. randomized (DBT=26; 1.11), 12mth 1.26 (SD 1.01) ES 0.32) vs TAU (Baseline Mean 1.11 (SD 0.82), 6mth 1.31 (SD 0.89). TAU TAU=16). DBT 12 mth 1.25 (SD 0.97) ES -0.17), ns. (n=41) participants: 11 CORE-OM (Functioning): No signif difference between DBT (Baseline Mean 2.63 (SD 0.83), 6mth completed 1 year of 2.49 (SD 0.83). 12mth 2.48 (SD 0.85) ES 0.18) vs TAU (Baseline Mean 2.51 (SD 0.61), 6mth 2.39 (SD treatment, 17 0.84), 12mth 2.22 (0.92) ES 0.47), ns. completed FU, 1 withdrew consent to use data. TAU CORE-OM (Wellbeing): No signif difference between DBT (Baseline Mean 3.02 (SD 0.95), 6mth participants: 14 2.79 (SD 1.08), 12mth 2.89 (SD 1.07) ES 0.14) vs TAU (Baseline Mean 3.0 (SD .70), 6mth 2.63 (SD completed FU. 1.07), 12mth 2.52 (SD 0.95) ES 0.69), ns. CORE-OM (Problems): No signif difference between DBT (Baseline Mean 2.89 (SD 0.64), 6mth 2.65 (SD 0.89), 12mth 2.62 (SD 0.97) ES 0.42) vs TAU (Baseline Mean 2.76 (SD 0.54), 6mth 2.53 (SD 0.90), 12mth 2.52 (SD .05) ES 0.44), ns. Secondary outcomes: NSSI (SASII): No signif difference between DBT (Baseline Mean 4.1 (SD 4.3), 6mth 2.9 (SD 5.0), 12mth 2.4 (SD 3.2) ES 0.40) vs TAU (Baseline Mean 7.8 (SD 4.7), 6mth 3.2 (SD 3.6), 12mth 3.1 (SD 3.4) ES 1.0), ns. Suicidal attempts (SASII): No signif difference between DBT (Baseline Mean 0.4 (SD .50), 6mth 0.3 (SD .47), 12mth 0.24 (SD .43) ES 0.32) vs TAU (Baseline Mean 0.5 (SD .51), 6mth 0.37 (SD .50), 12mth 0.06 (SD .25) ES 0.86), ns. Inpatient bed days (THI & medical record): No signif difference between DBT (Baseline Mean 1.1 71 (SD 3.0), 6mth 1.5 (SD 3.0), 12mth 1.7 (SD 3.6) ES -0.2) vs TAU (Baseline Mean 1.8 (SD 5.2), 6mth 2.1 (SD 5.4), 12mth 3.0 (SD 6.1) ES -0.23), ns Depression (BDI-II): No signif difference between DBT (Baseline Mean 35.2 (SD 9.7), 6mth 32.9 (SD 12.9), 12mth 32.2 (SD 12.8) ES 0.31) vs TAU (Baseline Mean 33.3 (SD 9.7), 6mth 29.1 (SD 12.4), 12mth 28.1 (SD 13.7) ES 0.54), ns Hjalmarsso 0, 6, 12 27 participants Primary outcomes: Outcomes reported for participants allocated to DBT for both intent-to-treat DBT = 19% n (2008) (post) included, 5 excluded (ITT) sample and treatment completers (COM). months due to treatment Global Assessment of Functioning Scale (GAF): Signif improvement in ITT sample (Baseline Mean Pre-post drop-out. 22 50.9 (SD 6.3), 6mth 59.9 (SD 8.0), 12mth 65.0 (SD 13.2), ES 2.24) p<.001; (n=27) participants included (COM Baseline Mean 51.4 (SD 6.3), 6mth 60.8 (SD 8.0), 12mth 67.4 (SD 13.4)) in analysis due to Depression (KABOSS-S): Signif improvement in ITT sample (Baseline Mean 26.3 (SD 9.2), 6mth 23.2 available data for (SD 12.4), 12mth 19.6 (SD 12.8) ES 0.73), p<.01; (COM Baseline Mean 26.3 (SD 9.2), 6mth 23.5 (SD those completing 12.0), 12mth 18.9 (SD 12.8)) treatment. Anxiety (KABOSS-S): No signif difference in ITT sample (Baseline Mean 25.2 (SD 8.9), 6mth 24.1 (SD 11.4), 12mth 21.8 (SD 11.0) ES 0.38, ns. (COM Baseline Mean 25.2 (SD 9.9), 6mth 23.7 (SD 11.0), 12mth 21.9 (SD 11.2)) Obsessive-compulsive (KABOSS-S): Signif improvement in ITT sample (Baseline Mean 23.4 (SD 7.5), 6mth 20.0 (SD 9.2), 12mth 17.7 (SD 9.5) ES 0.76, p<.01; (COM Baseline Mean 23.2 (SD 8.2), 6mth 20.4 (SD 8.6), 12mth 17.7 (SD 9.3)) Borderline (KABOSS-S): Signif improvement in ITT sample (Baseline Mean 33.5 (SD 9.1), 6mth 27.4 (SD 14.0), 12mth 23.2 (SD 13.8) ES 1.13; p<.001 (COM Baseline Mean 3.9 (SD 9.5), 6mth 28.1 (SD (SD 13.1), 12mth 23.0 (SD 14.0)) SCL-90-R (GSI): Signif improvement in ITT sample (Baseline Mean 2.1 (SD 0.6), 6mth 1.7 (SD 0.9), 12mth 1.5 (SD 0.9) ES 1) p<.01 (COM Baseline Mean 2.1 (SD 0.7), 6mth 1.6 (SD 0.8), 12mth 1.4 (SD 0.8)) SCL-90-R (PST): Signif improvement in ITT sample (Baseline Mean 69.9 (SD 11.6), 6mth 60.0 (SD 21.5), 12mth 58.4 (SD 20.7) ES 0.99, p<.05 (COM Baseline Mean 70.1 (SD 14.1), 6mth 60.2 (SD 20.8), 12mth 59.2 (SD 18.4)) SCL-90-R (PSDI): Signif improvement in ITT sample (Baseline Mean 2.7 (SD -.4), 6mth 2.3 (SD 0.7), 12mth 2.1 (SD 0.7) ES 1.5, p<.01 (COM Baseline Mean 2.7 (SD 0.5), 6mth 2.3 (SD 0.6), 12mth 2.1 (SD 0.6) ) Suicidal and NSSI (BPD-TOA & diary cards): Signif improvement in ITT sample (Baseline Mean 37.4 (SD 69.8), 6mth 9.4 (SD 18.7), 12mth 6.4 (SD 19.1) ES 0.44, p<.05 (COM Baseline Mean 15.4 (SD 72 13.5), 6mth 4.5 (SD 4.6), 12mth 1.4 (SD 2.1)) Results from DBT as percentages of improved and recovered: GSI: 8% improved, 23% recovered Depression (KABOSS-S): 20% improved, 20% recovered Anxiety (KABOSS-S): 20% improved, 7% recovered Obsessive-compulsive (KABOSS-S): 20% improved, 13% recovered Borderline (KABOSS-S): 27% improved, 27% recovered Priebe 177 participants Primary outcomes: DBT = 52% (2012) referred, 97 NSSI (interview): Means and SDs not provided. Participants in DBT condition showed significantly TAU = not excluded, 80 eligible greater reduction in self-harm over time. For every 2 mths in DBT, the risk of self-harm decreased reported Pragmatic to participate, 40 relative to TAU by an additional 9%. RCT of DBT allocated to Secondary outcomes: BPD symptoms (ZRSB): No signif difference between DBT (Baseline Mean vs. TAU treatment (DBT=4-; 17.9 (SD 6.8), 12mth 13.1 (SD 6.9) ES 0.71 ) vs TAU (Baseline Mean 18.4 (SD 7.6), 12mth 15.9 (SD (n=80) TAU=40). At FU: 10 7.5) ES 0.33), ns lost to FU, 7 BPD symptoms (BSI): No signif difference between DBT (Baseline Mean 122 (SD 41.2), 12mth uncontactable, 21 did 100.6 (SD 57.1) ES 0.52 ) vs TAU (Baseline Mean 134 (SD 39.3), 12mth 116 (SD 52.3) ES 0.46), ns not start or ended General Psychiatric Symptoms (BPRS): No signif difference between DBT (Baseline Mean 50 (SD prematurely. Total of 5.6)., 12mth 48 (SD 11.7) ES 0.36) vs TAU (Baseline Mean 52.8 (SD 9.9), 12mth 51 (SD 10.9) ES 0.18), ns 74 participants Quality of Life (MANSA): Signif improvement for DBT / No signif difference between DBT (Baseline included in analysis. Mean 3.3 (SD 0.8), 12mth 3.7 (SD 1.0) ES 0.5) vs TAU (Baseline Mean 3.2 (SD 0.9), 12mth 3.4 (SD 1.0) ES 0.22), ns Cost of care: Service costs higher for DBT participants than TAU participants (mean 5,685; SD 6,431 GBP; mean 3,754; SD 6,045 GBP, respectively). Despite higher cost, the difference was not significant. Estimated cost to achieve 1% point reduction in self-harm incidence through DBT is approximately 36 GBP. Stiglmayr 0, > 5 238 screened via Primary outcomes: DBT = 33% (2014) sessions, phone, 152 invited Number of suicide attempts (LPC): No signif difference between Baseline Median at baseline 0.00 4, 12 for assessment, 56 (range: 0-2), median at 12 mths 0.00 (range: 0-1) Pre-post (post) excluded, 18 referred NSSI (LPC): Signif improvement between Baseline Median at baseline 5.17 (range: 0-901), median (n=70) for CBT due to at 12 mths 1.00 (range: 0-174), p<.01 treatment group Number of inpatient stays (medical records): Signif improvement between Baseline 1.13 (SD 1.41), 73 capacity limitations, 12 mths 0.32 (SD 0.89), ES 0.57, p,.001 78 allocated to DBT. Psychiatric Hospital days (THI): Signif improvement between Baseline baseline 51.3 (SD 74.2), 12 Eight did not start mths 6.8 (SD 19.9), ES 0.60, p<.001 treatment, 17 Secondary outcomes: dropped-out during Borderline Symptoms (BSL): Signif improvement between Baseline Mean (2.10 (SD 0.54), 5 treatment, 6 declined sessions 1.89 (SD 0.73), 4 mths 1.81 (SD 0.74) and 12 mths (1.68 (SD 0.89), ES 0.78, p<.01 further participation Borderline Cognitions (QTF): Signif improvement between Baseline 3.73 (SD 0.55), 5 sessions 3.62 following treatment, (SD 0.55), 4 mths 3.42 (SD 0.71), 12 mths 3.25 (SD 0.82) , ES 0.87, p<.01 47 completed Depression (BDI): Signif improvement between Baseline (Mean 31.12 (SD 8.60), 5 sessions 26.78 treatment and (SD 12.24), 4 mths 24.27 (SD 10.22) and 12 mths (Mean 21.72 (SD 13.97)) ES 1.09, p<.001 provided data. Depression (HAM-D): Signif improvement between Baseline 11.69 (SD 4.67), 5 sessions 11.09 (SD 6.31), 4 mths 7.42 (SD 4.60), 12 mths 8.04 (6.11), ES 0.78, p<.05 General Psychiatric Symptoms (BSI-GSI): Signif improvement between Baseline 1.92 (SD 0.64), 5 sessions 1.95 (SD 0.73), 4 mths 1.64 (SD 0.80), 12 mths 1.46 (SD 0.90) , ES 0.72, p<.05 Dissociation-Tension-Scale (DSS): Signif improvement between Baseline 29.70 (SD 16.35), 5 sessions 25.95 (SD 19.08), 4 mths 22.27 (SD 17.26), 12 mths 20.02 (SD 16.84), ES 0.59, p<.01 Dialectical Behaviour Therapy studies of duration 6 months or less N.B. (only exception was Axelrod & Williams study which were 20 weeks) Axelrod 0, 10, 20 27 participants Primary outcomes: 44% (2011) (post) allocated to Depression (BDI): Signif improvement between Baseline Scores (25.07 (SD 9.77), 10 weeks 15.29 weeks treatment. Retention (9.34) and 20 weeks 14.5 (8.45) ES 1.08 p <.001. Pre-post rate not reported. Difficulties in emotion regulation (DERS): Signif improvement / No signif difference between (n=27) Baseline 118.00 (SD 18.47), 10 weeks 108.00 (SD 22.23), 20 weeks 94.80 (SD 17.89), ES 1.26, p < .001 Substance use: Signif improvement / No signif difference between Baseline 39.1% had weekly substance use in 30 days prior to the start of treatment. 8.6% had weekly substance use in 30 days prior to the end of treatment p<.01 Ben-Porath 0, 6 36 participants Primary outcomes: 12% 74 (2004) (post) enrolled, 10 Suicidal thoughts per month (Diary Card): Signif improvement between Baseline (Mean 12.71 (SD months excluded, 26 12.52) and 6 mths 7.82 (SD 8.04), ES 0.39), p<.05 Pre-post allocated to Parasuicidal behaviours (Diary Card): No signif difference between Baseline (Mean 11.43 (SD (n=26) treatment. 3 12.91), and 6 mths 9.71 (SD 11.32), ES 0.13) ns. *NB only 30% of sample identified as having dropped-out of parasuicidal behavior in the 6mths prior to treatment. treatment Secondary outcomes: prematurely General Functioning (SCL-90-R): No signif difference between Baseline (Mean 70.0 (SD 7.31) and (noncompliance). 23 6mth 65.8 (SD 9.31), ES = 0.57), ns. participants included Hopelessness (BHS): No signif difference between Baseline (Mean 10.57 (SD 7.15) and 6mth 10.04 in analysis. (SD 7.41), ES 0.07), ns. Psychiatric Symptoms (DPRS): Signif improvement / No signif difference between Baseline (Mean 5.23 (SD 1.79) and 6mth 3.57 (SD 1.57), ES 0.93), ns. Unemployment rates: Signif improvement / No signif difference between Baseline 78.3% unemployed and 6mth 60.9% unemployment, p<.01. Blenner- 0, 3, 6 21 participants Primary outcomes: 11% hassett (post) referred, 3 withdrew Number of self-harm episodes per week (Diary card): Signif not reported between Baseline (Mean (2009) 9 FU interest. 18 screened 3.75 (SD 4.50), 6 mths 1.00 (SD 2.14), ES 0.61, (N.B. only 4 self-harming at Baseline). months for eligibility, 4 Alcohol consumption units per week (Diary card): Signif not reported between Baseline (Mean Pre-post excluded, 2 withdrew 23.25 (SD 19.89), 6 mths 23.75 (SD 25.95), ES -0.03. (n=9) interest, 1 dropped- out prior to Over-the-counter doses per week (Diary card): Signif not reported between Baseline (Mean 40.88 treatment. Total of 9 (SD 71.94), 6 mths 22.75 (SD 51.06), ES 0.25. allocated to Psychiatric Hospital Days (medica records): Signif not reported between Baseline (Mean 57.83 and treatment, 1 6 mths 3.83) SD also not reported). dropped-out during General Psychiatric Symptoms (SCL-90-R - GSI): Signif improvement between Baseline and 6 mths treatment. Eight (Mean change 12.13 (SD 9.03), ES X), p <.05. included in analysis. General Psychiatric Symptoms (SCL-90R – PSDI): Signif improvement between Baseline and 6 mths (Mean change 18.00 (SD 14.50), ES X), p <.05 Risk (CORE): Signif improvement between Baseline and 6 mths (Mean change 1.21 (SD 0.71), ES 1.7,) p<.01. Problems (CORE): Signif improvement between Baseline and 6 mths (Mean change 1.52 (SD 0.65), ES 2.34) p<.01. Functioning (CORE): Signif improvement between Baseline and 6 mths (Mean change 1.06 (SD 0.40), ES 2.65) p<.01. 75 Wellbeing (CORE): Signif improvement between Baseline and 6 mths (Mean change 1.78 (SD 0.86), ES 2.1), p<.01. Brassing- 0, 6 11 participants Primary outcomes: 0% ton (2006) (post) allocated to Psychiatric Hospital Days (interview): Reduction Baseline (Mean M = 0.57 days per patient month months treatment, 1 prior to DBT; M = 0.2 days per patient per month during 6 mth of treatment (standard deviation Pre-post dropped-out within and significance not reported). (n=10) Although first month, 10 Borderline subscale (MCMI-III): Signif improvement between Baseline (Mean 90.50 (SD 15.66), 6 clinically included in analysis. mths 67.60 (SD 20.74) ES 1.46), p<.05 therapy Anxiety subscale (MCMI-III): Signif improvement between Baseline (Mean 92.70 (SD 10.40), 6 continue mths 74.30 (SD 19.53) ES 1.76), p<.05 d until 12 Depression subscale (MCMI-III): Signif improvement between Baseline (Mean 97.30 (SD 9.73), 6 months mths 50.30 (SD 35.51) ES 4.83), p<.001 NB – there was significant improvements across the following subscales for MCMI-III: borderline, paranoid, anxiety, depression, negativistic, somatoform, dysthymia, alcohol dependence, thought disorder, delusional disorder, disclosure and debasement (p <0.03 to p <0.0008). General Psychiatric Symptoms (SCL-90-R - GSI): Signif improvement between Baseline (Mean 2.18 (SD 0.18) and 6 mths 0.86 (SD 0.52), ES 7.3) p<.001. Positive Symptom Total (SCL-90-R - PST): Signif improvement Baseline (Mean 72.20 (SD 6.14); 6 mths 45.40 (SD 19.63) ES 4.36), p<.05 Positive Symptom Distress Total (SCL-90-R - PSD): Signif improvement / No signif difference between Baseline (Mean baseline 2.70 (SD 0.30), 6 mths 1.58 (SD 0.48), ES 3.73 p<.01 Carter 0, 3, 6 112 participants Primary outcomes: Number of self-harm episodes in previous 3 mth (LPC/PHI): No signif DBT = 47% (2010) (months) referred, 16 did not difference between DBT (Baseline Mean 22.00 (SD 28.60), 3mth 5.72 (SD 11.53), 6mth 5.27 (SD TAU = 11% attend assessment, 7.87) ES 0.58) vs WL+TAU (Baseline Mean 18.08 (SD 40.67), 3mth 6.13 (SD 11.41), 6mth 9.21 (SD (N.B. RCT of DBT Although 16 excluded, 1 with 31.22); ES 0.21), ns. People in vs. WL + clinically, no fixed address. 79 Secondary outcomes: Quality of Life Physical (WHOQOL-BREF): Signif improvement for DBT this TAU therapy eligible participants, (Baseline Mean 41.61 (SD 16.06), 3mth 52.85 (SD 19.52), 6mth 56.25 (SD 17.68) ES 0.91) vs WL+TAU (n=73) contin- 3 did not completed WL+TAU (Baseline Mean 40.70 (SD 22.35), 3mth 46.29 (SD 21.17), 6mth 42.39 (SD 21.46) ES 0.08), group ued until baselines p<.05 were 12 mths assessments Quality of Life Psychological (WHOQOL-BREF): Signif improvement for DBT (Baseline Mean 16.25 waiting to (excluded), 2 (SD 14.30), 3mth 31.04 (SD 21.26), 6mth 42.71 (SD 18.38) ES 1.81), vs WL+TAU (Baseline Mean commence withdrew consent 19.41 (SD 13.96), 3mth 29.72 (SD 22.77), 6mth 30.24 (SD 19.68)ES 0.78), p<.01 DBT) after assessments, 1 Quality of Life Social (WHOQOL-BREF): No signif difference between DBT (Baseline Mean 25.00 76 died by suicide (SD 18.53), 3mth 41.45 (SD 20.07), 6mth 48.75 (SD 19.36) ES 1.28), vs WL+TAU (Baseline Mean following baseline 36.55 (SD 23.73), 3mth 43.82 (SD 27.40), 6mth 49.73 (30.23) ES 0.27), ns. assessment, 3 not Quality of Life Environmental (WHOQOL-BREF): Signif improvement for DBT (Baseline Mean 47.34 included in analysis (SD 17.00), 3mth 57.81 (SD 13.69), 6mth 61.45 (SD 12.36) ES 0.83), vs WL+TAU (Baseline Mean due to missing data. 51.15 (SD 15.61), 3mth 56.30 (16.43), 6mth 54.57 (SD 18.16) ES 0.22), p<.05 76 randomized Days in bed (BDQ): Signif improvement for DBT (Baseline Mean 4.95 (SD 7.05), 3mth 3.00 (SD (DBT=38; TAU=35). 4.65), 12mth 2.70 (SD 4.28) ES 0.32), vs WL+TAU (Baseline Mean 3.96 (SD 7.02), 3mth 6.02 (SD 8.00), 6mth 7.29 (SD 8.33) ES -0.47) p<.05 Days out of role (BDQ): No signif difference between DBT 12.60 (SD 12.23), 3mth 8.68 (SD 9.79), 6mth 8.15 (SD 11.48) ES 0.36), vs WL+TAU (Baseline Mean 12.46 (SD 12.5), 3mth 11.36 (SD 11.42), 6mth 13.07 (SD 11.59) ES -0.05), ns. Pasieczny 0, 6 90 participants Primary outcomes: DBT = 7% (2011)3 (post) allocated to Suicidal attempts (medical record): Signif improvement between Baseline DBT (Mean 1.63 (SD TAU = 13% months treatment. 1.25) and 6 mths 0.28 (SD 0.68) ES 1.08), vs WL+TAU (Baseline Mean 1.66 (SD 1.64), and 6 mths DBT vs. WL 1.71 (SD 1.68) ES -0.03), p<.01. / TAU, but Although not clinically, 3 n.b. The self-report measures were not introduced until after 45 patients had been recruited, hence, the outcomes reported based on self-report measures are for n=45. randomise therapy NSSI (medical record): Signif improvement between Baseline DBT (Mean 13.60 (SD 18.99) and 6 d; based on contin- mths 3.15 (SD 5.79) ES 0.55), vs WL+TAU (Baseline Mean 19.07 (SD 25.34), and 6 mths 18.39 (SD group ued until 25.18) ES 0.03), p<.01. availability. 12 mths Psychiatric Hospital Days (medical record): Signif improvement between Baseline DBT (Mean (n=90) 13.15 (SD 18.22), and 6 mths 2.23 (SD 5.00) ES 0.60) WL+TAU (Baseline Mean 11.18 (SD 16.91), and 6 mths 13.6 (SD 17.84) ES -0.15), p<.01. ED visits (medical record): Signif improvement between Baseline DBT (Mean 3.38 (SD 5.01), and 6 mths 1.38 (SD 3.27) ES 0.40), vs WL+TAU (Baseline Mean 4.24 (SD 9.00), and 6 mths 3.76 (SD 7.81) ES 0.05), p<.01. Psychiatric admissions (medical record): Signif improvement between Baseline DBT (Mean 6mth pre-treatment 1.48 (SD 1.94), and 6 mths 0.48 (SD 1.11) ES 0.52), vs WL+TAU (Baseline Mean 6mth pre-treatment 1.46 (SD 2.01), and 6 mths 1.51 (SD 1.90) ES -0.02), p<.01. Secondary outcomes: 77 General Psychiatric Symptoms (BSI-GSI): Signif improvement between Baseline DBT (Mean 60.62 (SD 11.79), and 6 mths 51.52 (SD 13.62) ES 0.77), vs WL+TAU (Baseline Mean 60.07 (SD 12.94), and 6 mths 60.13 (SD 13.85) ES 0), p<.05 Depression (BDI-II): Signif improvement between Baseline DBT (Mean 34.91 (SD 13.03), and 6 mths 18.00 (SD 14.28) ES 1.3), vs WL+TAU (Baseline Mean 37.93 (SD 11.05), and 6 mths 33.93 (SD 13.86) ES 0.36), p<.01 Suicidal Ideation (BSS): No signif difference between Baseline DBT (Mean 14.43 (SD 9.97), and 6 mths 6.65 (SD 8.79) ES 0.78), vs WL+TAU (Baseline Mean baseline 15.20 (SD 11.12), and 6 mths 11.60 (SD 8.90) ES 0.32), ns. State Anxiety (STAI): Signif improvement between Baseline DBT (Mean 60.39 (SD 11.95), and 6 mths 49.57 (SD 16.05) ES 0.91), vs WL+TAU (Baseline Mean baseline 58.53 (SD 14.38), and 6 mths 57.20 (SD 13.71) ES 0.09), p<.05 Trait Anxiety (STAI): Signif improvement between Baseline DBT (Mean 64.0 (SD 10.42), and 6 mths 52.39 (SD 15.34) ES 1.11), vs WL+TAU (Baseline Mean 62.20 (SD 10.83), and 6 mths 61.87 (SD 11.55) ES 0.03), p<.01 Cost per patient: DBT average cost $12,196; TAU average cost $18,123 Prender- 0, 6 16 participants Primary outcomes: 31% gast (2007) (post) allocated to Parasuicidality (frequency) (Interview): No signif difference between Baseline (Mean 1.95 (SD months treatment . Five 2.91), 6 mths 1.85 (SD 3.22) ES 0.03), ns. participants dropped- Pre-post out: 1 due to Parasuicidality (medical severity) (Interview): Signif improvement between Baseline (Mean 2.30 (n=11) exacerbation of (SD 1.89), 6 mths 1.50 (SD 1.08) ES 0.42), p<.05 psychotic symptoms, Parasuicidality (intent) (Interview): Signif improvement between Baseline baseline 2.20 (SD 1.48), 2 due to need for 6 mths 1.45 (SD 1.07) ES 0.51), p<.05 hospitalization, 2 Psychiatric Hospital Days (medical record): Signif improvement between Baseline (Mean 6.09 (SD excluded as failure to 10.17), 6 mths 1.73 (SD 3.47). ES 0.43), p<.05 comply with Psychiatric Hospital Admissions (medical record): Signif improvement between Baseline (Mean treatment 0.72 (SD 1.01), 6 mths 0.27 (SD 0.47) ES 0.45), p<.05 requirements. Functioning (GAF): Signif improvement / No signif difference between Baseline (Mean 55.82 (SD 6.94), 6 mths 62.72 (SD 8.08) ES 0.99), p<.05 Depression (BDI): Signif improvement between Baseline (Mean 36.18 (SD 10.72), 6 mths 26.27 (SD 13.65) ES 0.92), p<.05 State Anger (STAXI-2): No signif difference between Baseline (Mean 72.64 (SD 27.65), 6 mths 75.36 (SD 20.41). ES -0.10), ns. 78 Trait Anger (STAXI-2): No signif difference between Baseline (Mean 79.18 (SD 19.52), 6 mths 78.36 (SD 17.28) ES 0.04), ns. Problem Coping (CSA): No signif difference between Baseline (Mean 54.97 (SD 15.69), 6 mths 57.16 (SD 11.73). ES 0.14), ns. Optimism Coping (CSA): Signif improvement between Baseline (Mean 50.64 (SD 11.65), 6 mths 55.91 (SD 12.88). ES 0.45), p<.05 Sharing Coping (CSA): Signif improvement between Baseline (Mean 44.32 (SD 25.10), 6 mths 55.91 (SD 16.86) ES 0.46), p<.05 Nonproductive Coping (CSA): Signif improvement between Baseline (Mean 72.82 (SD 7.97), 6 mths 68.43 (SD 9.19) ES 0.), p<.05 Face-to-face (number of contacts): No signif difference between Baseline (Mean 26.29 (SD 6.60, 6 mths 19.71 (SD 11.13). ES 1), ns. Face-to-face (duration of contacts): baseline 28.09 (SD 10.80), post 17.36 (SD 10.14) ES 1), p<.05 Telephone contact (number of contacts): Signif improvement / No signif difference between Baseline (Mean 11.43 (SD 12.84), 6 mths 7.86 (SD 6.07). ES 0.28), ns. Telephone contact (duration of contacts): Signif improvement / No signif difference between Baseline baseline 4.23 (SD 5.21), 6 mths 2.06 (SD 2.02) ES 0.42), ns. Williams 0, 20 140 participants Primary outcomes: DBT = 32% (2010) (post) attended a DBT skills Depression (BDI-II): No signif difference between Baseline DBT (Mean baseline 32.88 (SD 14.37), weeks training group, all post 25.24 (SD 13.03) ES 0.53; TAU baseline 37.01 (SD 14.06), post 24.81 (SD 17.84) ES 0.87), ns. TAU = 57% Pre-post allocated to Distress (K10+): No signif difference between Baseline DBT (Mean 34.83 (SD 11.84), post 22.67 (SD (n=31) treatment (individual 6.62) ES 1.32; TAU baseline 34.38 (SD 6.84), post 25.38 (SD 9.53) ES 1.03), ns. With DBT=31; TAU=109). Borderline Symptom Index (BSI): No signif difference between Baseline DBT (Mean 34.41 (SD compariso DBT participants: 21 7.27), post 24.88 (SD 9.28) ES 1.31; TAU 35.70 (SD 10.13), post 25.20 (SD 13.76) ES 1.04), ns. n with TAU completed DSM-IV screener (MSI-BPD): No signif difference between Baseline DBT (Mean 8.62 (SD 1.33), post (n=109) treatment, 10 non- 6.31 (SD 3.20) ES 1.74; TAU baseline 8.00 (SD 2.50), post 5.47 (SD 3.48) ES 1.01), ns. completers. TAU Quality of Life (BASIS-32): No signif difference between Baseline DBT (Mean 73.00 (SD 22.59), post participants: 47 49.47 (SD 21.65) ES 1.04; TAU baseline 75.37 (SD 26.77), post 54.57 (SD 32.60) ES 0.78), ns. completed treatment, 62 non- completers. 79 Conversational Model studies Stevenson 0, 12 85 participants Primary outcomes: Not (1992) (post) referred for BPD Criteria (SCID)/DSM-III severity score: Signif improvement between Baseline (Mean 17.40 (SD reported. assessment, 18 2.87)) and 12mth (Mean 10.50 (SD 5.08) ES 2.4), p<.001. Pre-post excluded. 67 met Self-harm episodes/year (Cornell): Signif improvement between Baseline (Mean 3.77 (SD 4.66)( (n=30) inclusion criteria, 8 and 12mth (Mean 0.83 (SD 1.18) ES 0.63), p<.001 excluded, 59 eligible Violent behavior episodes/year (Cornell): Signif improvement between Baseline (Mean 2.70 (SD for treatment, 11 4.05), 12mth 0.80 (SD 1.80) ES 0.47), p<.001 declined treatment Drugs used/day (Cornell): Signif improvement between Baseline (Mean 3.80 (SD 3.42), 12mth 0.63 or failed to attend. (SD 0.80) ES 0.93), p<.001 48 allocated to Medical visits/month (Cornell): Signif improvement between Baseline (Mean 3.50 (SD 2.75), treatment. Eight 12mth 0.47 (SD 0.57) ES 1.11), p<.001 participants dropped- Time away from work months/year (Cornell): Signif improvement between Baseline (Mean 4.47 out, 40 completed (SD 4.10), 12mth 1.37 (SD 2.57) ES 0.76), p<.001 treatment. Seven Hospital admissions/year (Cornell): Signif improvement between Baseline (Mean 1.77 (SD 1.52), continued ongoing 12mth 0.73 (SD 1.02) ES 0.68), p<.01 therapy, 33 Time as an inpatient months/year (Cornell): Signif improvement between Baseline (Mean 2.87 (SD completers. Three 2.33), 12mth 1.47 (SD 1.87) ES 0.60), p<.05 did not complete 12mth FU, 30 Cornell index score: Signif improvement between Baseline (Mean 42.63 (SD 14.90), 12mth 28.63 included in analysis. (SD 13.35) ES 0.94), p<.001 Korner 0, 12 29 participants Primary outcomes: Not (2006) (post) entered treatment – BPD Criteria (SCID): Signif improvement between CM (Baseline Mean 14.86 (SD 3.61), 12mth 9.24 reported. no details provided (SD 6.15) ES 1.56 vs WL baseline 14.27 (SD 3.9), 12mth 13 (SD 3.02) ES 0.33), p<.001. Un- about recruitment. Functioning (GAF): Signif improvement between CM (Baseline Mean 48 (SD 13.7), 12mth 64.56 controlled Compared with 31 on (SD 16.7) ES 1.21 vs WL baseline 44.85 (SD 11.2), 12mth 50.56 (SD 13.7) ES 0.51), p<.001 WL + TAU the WL for (n=60) treatment. Effect sizes for DBT and CM outcomes are shown for within treatment. Effect sizes are also shown within treatment for the studies that had a comparison group. The significance level is as reported in the paper, hence, for pre-post studies refers to significant change over time. Where there is a comparison group, significance tests are shown for time by group interaction. Effect sizes are most straightforwardly interpreted as one of three categories: small (0.2), moderate (0.5), and large (0.8). It is noted that effect sizes are not ideal for non-normally distributed data, however, they are acceptable and were used in this chapter in order to have a standardised 80 number with which to compare outcomes across studies. Effect sizes were calculated by calculating the difference between the Means at Pre and Post-treatment and dividing by the SD at Pre-treatment. COM = Treatment Completers; CM = The Conversational Model; DBT = Dialectical Behaviour Therapy; ITT = Intent-to-treat; RCT = Randomised Controlled Trial / Randomised Clinical Trial; TAU = Treatment as Usual; WL = Waitlist. Outcome measures: BASIS-32 = Behaviour and Symptom Identification Scale; BDI = Beck Depression Inventory I; BDI-II = Beck Depression Inventory II; BDQ = Brief Disability Questionnaire; BSI-GSI = Brief Symptom Inventory – Global Severity Index; BSI = Borderline Symptom Inventory; BSL = Borderline Symptom List; BSS = Beck Scale for Suicidal Ideation; CORE-OM = Clinical Outcomes in Routine Evaluation – outcome measurement; DERS = Difficulties in Emotion Regulation Scale; DPRS = Derogatis Psychiatric Rating Scale; DSS = Dissociation-Tension-Scale; GAF = Global Assessment of Functioning; GSI = Global Severity Index; HAM-D = Hamilton- Depression Scale; KABOSS-S = Karolinska Affective and Borderline Symptom Scale – Self-Assessment; K10+ = Kessler 10 Scale; LPC = Lifetime Parasuicide Count; MANSA = Manchester Short Assessment of Quality of Life; MCMI-III = Millon Clinical Multiaxial Inventory, Third Edition; MSI-BPD = McLean Screening Instrument for Borderline Personality Disorder; QTF = Questionnaire of Thoughts and Feelings; SASII = Suicide Attempt Self Injury Interview; SCID-BPD = Structured Clinical Interview for DSM-IV, Personality Disorders, Borderline Personality Disorder criteria; SCL-90-R = Symptom Checklist 90 Revised; BPRS = Brief Psychiatric Rating Scale; STAI = State Trait Anxiety Inventory; WHOQOL-BREF = World Health Organization Quality of Life Assessment – Brief form; ZRSB = Zanarini Rating Scale for Borderline Personality Disorder. 1.6.1 DBT Effectiveness Research There were five published DBT studies that investigated DBT over 12 months, the length of time the treatment was originally designed to be delivered. These studies had marked variability in the sample sizes, ranging from 27 to 80 participants. While DBT was originally developed as a 12-month treatment, and all but one of the efficacy studies of DBT were for 12 months, many of the effectiveness studies only investigated DBT for six months. There were eight published DBT studies that investigated Standard DBT for six months or less duration, and in these studies, there was even greater variability in the sample sizes ranging from nine to 90 participants. The effectiveness studies described in this section were conducted in a number of Western countries (the United Kingdom, Ireland, Sweden, Germany, the United States, New Zealand, and Australia). The profile of the samples was fairly similar across the studies, including a formal diagnosis of BPD and age range of 18–65 years. Suicidal and NSSI behaviour were not among the inclusion criteria for all studies, which makes it difficult to compare across studies in terms of suicidal outcomes. Further, there was marked variability in outcome measures used across the studies, which further limits the ability to compare studies. Exclusion criteria were fairly consistent across the studies and were few in number, the most common being mental retardation and/or substance dependence. Some studies had tighter exclusion criteria, e.g., Carter et al. (2010) while others were more open as to who they accepted, e.g., Comtois et al. (2007). Hence, in a sense, the profile of participants included in these samples was fairly homogenous in age and diagnostic criteria but differed regarding their severity. Four effectiveness studies have compared DBT with a control group of either TAU or Waiting List + TAU (Carter et al., 2010; Feigenbaum et al., 2012; Pasieczny & 81 Connor, 2011; Priebe et al., 2012). In each of these studies, suicidal and NSSI behaviour was significantly reduced among both DBT participants and participants in the control group. However, Pasieczny and Connor (2011) and Priebe et al. (2012) were the only studies to find a significant difference between those receiving DBT and the control group in the reduction of suicidal and non-suicidal self-injury. Pre-post studies of outcomes of DBT in routine clinical settings have been conducted in a number of community mental health settings. The clientele and structure of mental health settings are quite variable in different parts of the world, which makes it hard to compare from one setting to another. In studies that examined pre-post scores for those who had completed DBT, most found either a significant and/or clinically significant reduction in suicidal and NSSI scores. However, in many of the studies not all of the participants were engaging in suicidal and NSSI at the start of treatment (Blennerhassett, Bamford, Whelan, Jamieson, & O’Raghaillaigh, 2009; Brassington & Krawitz, 2006; Hjalmarsson, Kaver, Perseius, Cederberg, & Ghaderi, 2008; Stiglmayr et al., 2014). In Ben-Porath, Peterson, and Smee (2004) only 30% of the sample reported a history of NSSI in the six months preceding the study and the reduction in their self-harming behaviour across the six months was not significant. Most of the studies focused on reductions in suicidal and NSSI rather than changes in severity. Prendergast and McCausland (2007) reported that participants did not change markedly in terms of reduction of self-harming behaviour. However, 45% of their sample had a significant reduction in the severity of self-harming incidents, such that all of the self-harming behaviour occurring at post-treatment met their criteria of “no danger,” e.g., scratching. Of note is that a number of these studies measured NSSI and suicidal behaviour by self-report on a diary card. It is not clear how 82 reliable this was as a source of data as participants may under-report, over-report, or accurately report, and as such, there may have been a response bias. There was a large range in dropout rates from therapy between the different effectiveness studies ranging from 0% of 10 participants (Brassington & Krawitz, 2006) to 56% of 41 participants (Feigenbaum et al., 2012). It is difficult to speculate about the variability in dropout rates across studies, given that the studies were from a range of different contexts with slightly different inclusion and exclusion criteria. Regardless, the results of effectiveness studies show us that a substantial portion of people drop out of DBT treatment in routine clinical settings and while there has been some work in understanding this, the reasons for it are largely unclear. Many of these studies only had data for those who completed treatment, not those who dropped out. Hence, we do not know whether those who dropped out showed similar improvement to those that stayed in therapy or whether they had a poorer outcome, and the wide range of variability rates skews the picture of reported outcomes. Amongst studies that compared DBT with a control group (Carter et al., 2010; Feigenbaum et al., 2012; Pasieczny & Connor, 2011; Priebe et al., 2012), there was consistency in changes on the primary outcomes. However, the secondary outcomes varied, with some overlaps and some distinct constructs assessed. Of the secondary outcomes measured, three studies (Carter et al., 2010; Feigenbaum et al., 2012; Priebe et al., 2012) found largely no significant difference in DBT when compared with the control group, with some minor exceptions: Carter et al. (2010) found a significant difference in favour of DBT for reduction in bed days and quality of life, and Priebe et al. (2012) showed that those in DBT had significantly fewer days in hospital by the end of 12 months. However, in addition to showing a significant reduction in self-harming 83 behaviour, Pasieczny and Connor (2011) found significant results on all secondary outcomes. It is not clear why they found significant results on the secondary outcomes while the other studies did not. As discussed above, one possibility was that the TAU in their setting was less satisfactory than in the other studies. Across the effectiveness studies, there was inconsistency regarding the outcomes. The pattern of results was more consistent within studies than between studies, such that some studies found very few significant findings across their outcomes while others found a range of significant outcomes. There are a number of possible interpretations; while the inclusion criteria were fairly homogenous, there may have been a different profile of patients in terms of their severity and comorbidities between services. Alternately, it may be that different aspects of the treatment were emphasised in different settings. Regardless, the overall picture is that treatment with DBT was associated with a reduction in NSSI and suicidal behaviour, days admitted to psychiatric hospital, depression severity and general psychiatric symptoms, and improvements in quality of life. There was not a clear pattern of outcomes between those studies with smaller samples versus larger samples. The level of improvement in these domains varied between studies, however, and possible reasons for this variation are worth considering. In the studies that incorporated a control or comparative condition, the alternate condition also showed improvement, but none of the studies have large enough sample sizes to detect whether DBT was superior or equal to the control or comparative condition. It is possible that the DBT being delivered in the effectiveness trials was inferior to the quality in the RCTs, and further effort is required to improve the quality of DBT in routine clinical settings. As many of these studies did not include adherence ratings, 84 we do not have information about whether the DBT delivered was adherent. Furthermore, some studies accepted more psychiatrically disabled persons into their treatments, e.g., Comtois et al. (2007) and Priebe et al. (2012), who had very few exclusion criteria. There are several limitations in the above studies, with eight of the studies not utilising a control group, and hence, it is not possible to attribute the gains to the specific DBT intervention. The majority of studies also had small sample sizes, which limits generalisability. 1.6.2 CM Effectiveness Research The research in CM is substantially more limited in comparison with DBT. CM has been evaluated in two observational studies: a pre-post design with five-year follow-up, followed by a study involving a waiting list control group. The treatment developer, Russell Meares, and his team in Sydney, Australia, conducted the existing studies. The outcomes from the observational studies were promising. As seen in Table 1.5, the first study by Stevenson and Meares (1992) reported on 30 patients who completed a year of treatment, comparing their scores for the year before commencing therapy with the year of therapy. A total of 48 patients consented to treatment, however, eight patients dropped out in the first months and 10 others were not included in the outcome analysis (seven continued therapy and three were unable to be contacted). Participants who received the year of CM showed significantly reduced rates of suicidal and violent behaviour, drug use, time away from work, and reduced medical visits. About one-third (30%) of participants no longer met the criteria for BPD. These results were maintained at five-year follow-up (Stevenson et al., 2005). The study reported only on patients who had completed therapy and, as such, who we 85 may expect to have done well. In addition, the comparison was based on using the patients as their own controls, so that the comparison data was from the patients’ and their loved ones’ self-reports of their year prior to therapy. Despite the lack of randomisation and a separate control group, this was one of the first reports demonstrating that BPD patients could improve and be treated successfully. In addition, unlike many other outcome studies in BPD that ignore the issue of medication, all patients were gradually withdrawn from medication. Hence, the improvement cannot be attributed to pharmacological benefit. Some years later, Meares et al. (1999) utilised an opportunity of forming a natural control group with people who had been on the waiting list for a minimum of 12 months to serve as a post-hoc comparison group. They compared the treated cohort described above with a group of people later referred to the service. The treated group and the wait list group were similar in their demographics. After a year of treatment, there was a significant difference between those who had received a year of treatment of CM and those who had been on the waiting list for a year. Comparing the treated group with a waiting list control is stronger methodologically than the previous study, which involved using the patients as their own controls and eliciting retrospective self-reports of those patients. However, there was a significant time lag between these studies and while the actual time lag is not specified in the articles, there were seven years between publications, and therefore, the treated group and the waiting list control group are temporally different. Stevenson et al. (2005) published the results of the five-year follow-up. Impressively, they managed to locate all 30 patients five years later and hence, there was no missing data to account for. Scores on all measures showed that the gains had been maintained at five years 86 and a slightly higher percentage of 40% no longer met the criteria for BPD. The authors noted that most of the participants had some ongoing professional contact and consequently it is not possible to attribute the maintenance of the gains to the original treatment. Furthermore, the assessments were not conducted by a blind assessor, but instead by the principal investigator and first author of the study. The authors noted that it was for this reason that all patients cooperated with the follow-up. Korner et al. (2006) followed up the original Stevenson and Meares (1992) study with a similar sample size of 29, but this time with a waiting list control group that received TAU whereas the treated cohort were receiving one year of therapy. The rate and degree of improvement for the treated group in this sample was comparable to the treated group in the original study. The research was conducted when the clinic had been set up for more than a decade, so on the one hand could be seen as usual clinical conditions. On the other hand, the treatment developer remained embedded in that clinic, so the research is not an independent replication. The issue of fidelity to treatment across all of the CM studies was addressed in the following ways: all therapy sessions were audiotaped and reviewed with a supervisor weekly. If there were concerns about adherence to the model, the supervisor could provide corrective feedback. However, it is a weakness that systematic objective adherence coding of sessions was not conducted. As CM has not yet been tested in a randomised trial, it is not possible to conclude that it is superior to treatment as usual. Randomisation procedures are used to ensure that the treatment and control groups are comparable at the pre-treatment time point before any individual receives the intervention (Howard et al., 1996). CM has also not been evaluated against another evidence-based treatment for BPD, and 87 consequently, there is no way of knowing whether it achieves similar outcomes as treatments with significantly greater evidence base. While adherence coding was not conducted, from the authors’ point of view, monitoring of fidelity was built into the supervision with the expert supervisors listening to audiotapes and giving immediate feedback. Like many of the studies in the field of BPD, the sample sizes in the CM studies were relatively small. 1.7 Rationale for the Thesis The field has made much progress in the past 30 years in the development and evaluation of treatments specifically for BPD. As outlined in this chapter, we now know that it is possible for BPD to be successfully treated. However, much further research remains to be done. 1.7.1 Gaps in the Literature Overall, the studies published so far in the area of outpatient psychotherapeutic treatment for BPD have several limitations. Many have small sample sizes (range n = 10 to 180), the majority have not reported power calculations and as such, are likely to be underpowered to detect significant differences. Apart from DBT, most treatment models have only been evaluated in one or two studies. The majority of studies have been conducted by treatment developers or investigators that are strongly allegiant to a particular model. Where studies involved all of the investigators with allegiance to one model, outcomes have been consistent in support of that model (Luborsky et al., 1999). Researcher allegiance is now well recognised as affecting results (Leichsenring et al., 2017). In the most recent Cochrane review of psychological therapies for BPD, Stoffers et al. (2012) suggest “there is an urgent need 88 for independent research endeavours” (p. 77). Hence, replications are needed, particularly by researchers separate to the treatment developers. Most studies have been compared against TAU and/or waiting list. It is well established that passive control conditions can lead to an overestimation of the efficacy of a psychotherapy (Mohr et el., 2014) and that patients allocated to waitlist controls often experience worsening of symptoms (Furukawa et al., 2014). Hence, a positive finding for the active treatment is not surprising. There is no evidence from comparisons of active models developed specifically for the treatment of BPD that any one form of psychotherapy is superior to any other. Further, many of the existing studies were concerning in terms of their methodological quality. Risk of bias was assessed using the Cochrane Collaboration’s Risk of Bias tool (Higgins & Green, 2011). Each item was evaluated as being high, low or unclear risk as per the criteria provided in the Cochrane Handbook for Systematic Review of Interventions (Higgins & Green, 2011) and an overall risk of bias calculated. Table 1.7 shows the assessment of risk of bias across the studies discussed in Chapter One. All non-RCTs were automatically rated as high overall risk of bias. The item regarding blinding of therapists and participants is clearly not possible in psychotherapy trials. Hence, consistent with published concerns (Baker et al, 2012), this item was disregarded. Eight (40%) of the efficacy studies were identified as high risk, six (30%) were identified as medium risk and six (30%) were identified as low risk. In contrast, 12 (80%) of the effectiveness studies had high risk, 3 (20%) had medium risk and none (0%) were identified as low risk. Hence, it is clear that the existing effectiveness studies have might higher risk of bias than the efficacy studies. As such, results of significant findings need to be interpreted with a degree of caution. 89 The majority of the research in psychotherapy for BPD consists of efficacy studies set in highly controlled settings, or at the other end, comprises small pre-post studies in routine clinical settings. McMain, Newman, Segal, and DeRubeis (2015) assert that effectiveness studies have limitations and that there are weaknesses in the quality of many of these studies. With studies conducted in university settings by highly trained therapists, it is unclear how well the effects generalise to real world clinical settings (Roy-Byrne et al., 2003). As can be seen in Table 1.7, the efficacy studies all have substantially less risk of bias than the effectiveness studies. There is a need for methodologically high quality, well-powered studies to be conducted in routine clinical settings to see how well the findings of the efficacy research hold up in those settings. Of the existing DBT effectiveness studies, only a minority have used some form of adherence coding to establish fidelity to the model. None of the CM studies had adherence coding. Adherence is usually collected in RCTs where the range of adherence is most likely both higher and narrower than in routine clinical practice. Hence, for many of the studies, it is unclear to what extent either of the treatments, as developed, were being delivered. While the authors intended to measure the impact of DBT or CM on their participant groups, we do not know much about the DBT or CM that was delivered. Across BPD studies in general, most studies have focused on suicidal and NSSI as the primary outcome. Few studies have reported on whether patients have made clinically significant change or continue to meet criteria for BPD following treatment. Amongst published studies, many lack significant follow-up beyond the end of 90 treatment (Cristea, 2017) and consequently, there is a gap in our knowledge regarding how well the results of the treatment are maintained. 91 Table 1.7 Assessment of Risk of Bias of Studies Study Random Allocation Blinding Blinding Absence of Free of Adherence Summary of sequence concealment (participants (outcome incomplete selective to protocol risk of bias generation & personnel) assessment) outcome reporting data BPD Efficacy studies Bateman (1999) Yes Unclear No Unclear No Yes Yes HIGH Bateman (2009) Yes Yes No Yes Yes Yes Yes LOW Blum (2008) Yes Yes No Unclear No Yes Yes MEDIUM Bos (2010) Yes Unclear No Unclear Yes No No HIGH Clarkin (2007) Yes Unclear No Unclear Unclear No Unclear HIGH Cottraux (2009) Yes Yes No Yes Yes No Unclear MEDIUM 92 Davidson (20006) Yes Yes No Yes Yes Yes Yes HIGH Doering (2010) Yes Yes No Yes Yes Yes Yes LOW Farrell (2009) Yes Unclear No Unclear No Yes No HIGH Giesen-Bloo (2006) Yes Yes No Yes Yes Unclear Yes LOW Gratz (2006) Yes Unclear No No Unclear Yes Yes HIGH Gratz (2014) Yes Unclear No Yes Yes Yes Yes MEDIUM Jorgensen (2012) Yes Unclear No Unclear Yes Yes No HIGH Koons (2001) Yes Unclear No Unclear Yes Yes Yes MEDIUM Linehan (1991) Yes Unclear No Unclear No No Yes HIGH Linehan (1999) Yes Unclear No Yes Yes No Yes MEDIUM Linehan (2002) Yes Unclear No Yes Yes No Yes MEDIUM Linehan (2006) Yes Yes No Yes No Yes Yes LOW McMain (2009) Yes Yes No Yes Yes Yes Yes LOW Verheul (2003) Yes Unclear No Unclear No No Yes HIGH DBT Effectiveness studies of 12 months Comtois et al. (2007) No No No No No Yes No HIGH Feigenbaum et al. (2012) Yes Yes No Unclear Yes Yes No MEDIUM Hjalmarsson (2008) No No No No Yes Yes No HIGH Priebe (2012) Yes Yes No Unclear Yes Yes Yes MEDIUM Stiglmayr (2014) No No No No Yes Yes Yes HIGH DBT Effectiveness studies of 6 months or less Axelrod (2011) No No No No Yes Unclear No HIGH Ben-Porath (2004) No No No No No Yes No HIGH Blennerhassett (2009) No No No No No Yes No HIGH Brassington (2006) No No No No Yes Yes No HIGH Carter (2010) Yes Yes No No Yes Yes No MEDIUM Pasieczny (2011) No No No No Yes Yes Yes HIGH Prendergast (2007) No No No No No Yes No HIGH Williams (2010) No No No No No Yes No HIGH Current study Yes Yes No Yes Yes Yes Yes LOW CM Effectiveness studies Stevenson No No No Yes Yes Yes No HIGH 93 Korner No No No Unclear Unclear Yes No HIGH According to the Cochrane Collaboration Handbook: Sequence generation: Was the allocation sequence adequately generated? Were the methods used to generate the allocation sequence clearly described? Allocation concealment: Was allocation adequately concealed, so that it was not possible to determine allocations before or during enrolment? Blinding of participants and personnel: Was knowledge of the allocated intervention adequately prevented during the study from patients and therapists? Blinding of outcome assessors: Was knowledge of the allocated intervention adequately prevented during the study from the outcome assessors? Incomplete outcome data and withdrawals: Were intention to treat analyses performed? Had participants withdrawn from the study? Free of selected reporting: Were all pre-specified and expected outcomes of interest reported? Adherence to protocol: Was there a description of methods used to assess adherence and of adherence standards? Yes = low risk of bias; No = high risk of bias; Unclear = unclear risk of bias Summary of risk of bias: Low (6 items with low risk of bias), Medium (4-5 items with low risk of bias) and High (1-3 items with low risk of bias). N.B. All non-RCTs were automatically rated as high overall risk of bias. 1.7.2 How the Current Study Will Address the Gaps The present study comparing DBT with CM attempts to address some of the limitations identified above. DBT has mostly been compared against TAU instead of an active treatment comparison (Stoffers et al., 2012). This study is adequately powered with a sample size of 162. It utilises an RCT design which means that outcomes can be attributed with greater confidence to the impact of the distinct treatments that are being compared by minimising sources of bias (Midgley, Ansaldo, & Target, 2013). Randomised clinical trials are recognised as the gold-standard of evidence-based medicine. However, comparative trials of two psychotherapies, also labelled as ‘horse race’ studies have been criticised in that meta-analyses have not revealed substantial differences between the therapies (Locher, Gaab, & Blease, 2018). This is consistent with the ‘Dodo bird verdict’ put forward by Saul Rozenzweig (1936) and since supported many times (see Luborsky et al., 2002) that posits that when active psychotherapies are compared, there will only be small differences. CM is taught and used in Australia and has studies with promising outcomes (Korner & McLean, 2017), but has not been evaluated in a randomised trial. Both treatments have been specifically developed to treat BPD; however, the two treatments have different theoretical underpinnings about the therapeutic mechanisms. DBT is a behavioural therapy that actively targets reduction of self- destructive behaviour, whereas CM is a psychodynamic therapy that targets development of a healthy sense of self. As such, we would expect that they might work differentially in the outcomes they achieve. Whilst the current study could be thought of as a ‘horse race’ study, instead we are looking at whether DBT performs superior to CM in reducing self-destructive behaviour and whether CM performs superior to DBT 94 on a proxy for sense of self. Hence, the question is whether both so-called ‘horses’ might win but on different domains. The researchers in the present study were not involved in development of the treatments, nor do they work alongside the treatment developers. Most importantly, the principal investigators have allegiance to both therapeutic models, hence, reducing some of the bias that comes from researcher allegiance in ‘horse race’ studies and the study was carried out in a real world setting with clinicians employed in public sector mental health services, who are usually responsible for implementing these therapies. The current study is included in Table 1.7 to show the ways that risk of bias has been reduced in our study: by use of adequate random sequence generation, allocated concealment procedures, assessors blind to intervention status, complete outcome data using intention to treat analyses, all pre-specified outcomes reported and adherence to protocol measured and reported. The study has been reported utilising the Consolidated Standards of Reporting Trials (CONSORT; Schulz et al., 2010) checklist in order to ensure complete, clear and transparent information regarding the methodology and findings. 1.8 Thesis Aims This thesis comprises seven chapters that describe the methodology of a RCT conducted in a public sector mental health service comparing DBT and CM in the treatment of BPD and the results of that trial. The thesis aims to contribute to knowledge and practice regarding psychotherapeutic treatment of BPD. The aims are outlined here and the chapters in which they will be addressed are described in more detail below in the overview of the thesis. 95 1. Develop a psychometrically valid measure to test adherence to CM (Chapter 3). 2. Evaluate the effectiveness of CM against DBT, in a routine clinical setting (Methodology in Chapter 2; Outcomes in Chapter 4). 3. Compare CM and DBT in relation to the working alliance (Chapter 5). 4. Explore improvement and deterioration at the individual patient level (Chapter 6). 1.9 Overview of the Thesis This thesis consists primarily of the development and evaluation of an RCT for BPD in a routine clinical setting. Chapter 2 describes the methodology of the RCT. Assessing fidelity to treatment is important in an RCT, to be confident that the therapy has been delivered adherently (Vaughan & Ochoa, 2016). A tool for assessing adherence to DBT was already in existence and was utilised in this study; however, no such tool was available for CM. Chapter 3 describes the development and evaluation of the Newcastle Adherence Scale for Conversational Model (NASCOM), inter-rater reliability of the measure, and its capacity to discriminate between the two therapies. Severity of depression scores and suicidal and NSSI were identified as the primary outcomes of the RCT. Secondary outcomes included BPD severity, dissociation, interpersonal problems, difficulties with the sense of self, and mindfulness and emotion regulation capacity. Chapter 4 reports on the findings of those primary and secondary outcomes in the RCT. Chapter 5 reports on the comparison of therapeutic alliance in DBT and CM, for therapist-rated and client-rated alliance as well as distinct components of the alliance in the task, goal, and bond. 96 In addition to aggregated results by treatment group, it is important to explore clinical improvement and deterioration at an individual patient level. Chapter 6 focuses on recovery for BPD patients and individual patterns of improvement and deterioration. Finally, Chapter 7 discusses the overall findings from this body of work, highlights important conclusions, and discusses next steps for research. 97 Chapter 2 Methodology of a Randomised Clinical Trial of Dialectical Behaviour Therapy and Conversational Model Chapter 1 outlined the published psychotherapy research to date of outcome studies for BPD and provided the rationale and aims of the current study. Chapter 2 outlines the methodology of the RCT to provide clarity regarding the design of the study, setting, participant sample, treatments involved, and the plan for the data analysis. Results from the study will be reported in Chapters 4–6. 98 Borderline Personality Disorder (BPD) is associated with a high degree of suffering, high rates of suicide attempts (Lieb et al., 2004), and a lifetime suicide mortality rate of approximately 10% (Black et al., 2004). There is a high rate of comorbidity with Axis I disorders, which has been estimated at 85% (Leichsenring et al., 2011). Beyond high levels of symptomatic impairment, large-scale studies have shown pervasive social and functional impairment (Gunderson et al., 2011). Community BPD prevalence rates are estimated to be in the range of 1–4%. This is higher within mental health settings, where approximately 10% of all psychiatric outpatients and between 15 and 25% of psychiatric inpatients meet the criteria for BPD (Leichsenring et al., 2011). It is commonly acknowledged that BPD is a challenging disorder to treat (Leichsenring et al., 2011) and one that evokes high stigma amongst mental health practitioners (Aviram et al., 2006). In the past 20 years, there has been considerable progress in treatments developed and evaluated for BPD. Psychiatric medication is, at best, of modest benefit over a short period and psychotherapy is the indicated treatment for BPD (NHMRC, 2012). Dialectical Behaviour Therapy (DBT) is the most intensely studied psychotherapy for the treatment of BPD (Stoffers et al., 2012). DBT is an adapted form of Cognitive Behaviour Therapy specifically developed for women with BPD who are engaging in suicidal and non-suicidal self-injurious behaviours. DBT has been the focus of more clinical trials than any other psychotherapy for BPD. Across these trials, results generally show that treatment with DBT is associated with a reduction in suicidal and non-suicidal self-injurious behaviours, inappropriate anger, and service utilisation. 99 Two studies have compared DBT with an active treatment. Linehan et al. (2006b) compared DBT with “treatment by experts,” where those identified in the community as having expertise in treating BPD using treatments other than DBT, provided treatment to the control group. DBT was found to be superior on outcomes of suicidal behaviour, treatment retention and service utilisation. McMain et al. (2009) compared DBT with “Generalised Psychiatric Management” and found that both treatments demonstrated significant reductions in suicidal and self-harming behaviour, as well as on a range of other clinically relevant measures. There was no significant difference between the two treatments. In addition to DBT, a number of other forms of Cognitive Behaviour Therapy appear promising for the treatment of BPD (e.g., Systems Training for Emotional Predictability and Problem Solving, Schema Focused Therapy and Cognitive Therapy), although more studies are needed (Stoffers et al., 2012). These have been described in Chapter 1. Beyond cognitive behavioural therapies, there is empirical evidence for a number of psychodynamic therapies, namely Mentalisation Based Therapy (Bateman & Fonagy, 1999, 2009), Transference Based Therapy (Clarkin et al., 2007) and the American Psychiatric Association’s General Psychiatric Management (McMain et al., 2009). All have shown positive results in the treatment of BPD, including reductions in suicidal and non-suicidal self-injurious behaviour, as well as general functioning. There is no direct evidence from comparisons of active treatments developed specifically for BPD that any one form of psychotherapy is superior to any other. The Conversational Model (CM) is a psychodynamic model of treatment developed specifically for BPD and targets the development of a healthy sense of self (Meares, 2005). It focuses heavily on the therapeutic relationship as a template for other 100 relationships and aims to help individuals increase their capacity for reflective functioning (Meares, 2004). Although CM was developed specifically for the treatment of BPD, it has not yet been tested in a randomised trial. As was described in Chapter 1, it has been evaluated in a published study (Stevenson & Meares, 1992), using a pre- post design, and a replication study (Korner et al., 2006). Both studies showed significant reductions in the number of BPD diagnostic criteria met and reduction in suicidal behaviour and non-suicidal self-injurious behaviour, violent behaviour, and hospital admissions. Participants in the initial trial were followed up after five years and the gains were found to be maintained (Stevenson et al., 2005). The aims of the current study were to evaluate DBT in a routine clinical setting and to compare CM against another therapy for BPD with an established evidence base. 2.1 The Current Study The present study compared DBT with CM and attempted to address the limitations of previous research. The study has a large sample size, utilises intention to treat analyses, and has incorporated measures of fidelity to treatment. The researchers were not involved in the development of either treatment nor do they work alongside the treatment developers. Most importantly, the principal investigators have allegiance to both therapeutic models, and the study was carried out in a real world setting with clinicians employed in public sector mental health services, who are responsible for implementing these therapies. Fidelity was given close attention to ascertain how adherent clinicians in a real world setting were to the treatment protocol. 101 DBT and CM have different theoretical underpinnings. While DBT is a Behavioural Therapy that actively targets reduction of self-destructive behaviour and an increased capacity for emotional regulation (Linehan, 1993a), CM is a psychodynamic therapy that targets an internal sense of well-being and development of a healthy sense of self (Meares, 2004). As such, we would have expected that they might be differentially effective. 2.2 Methods/Design 2.2.1 Study Design and Aims This was a single site, two-armed parallel RCT designed to investigate the effectiveness of CM and DBT in a public sector mental health service in Australia. The sample population comprised adults with a primary diagnosis of BPD and recent suicide attempts and/or non-suicidal self-injury (NSSI) episodes. The main aim of this study was to compare CM with DBT for two co-primary outcomes at post-treatment (14 months): change in the number of combined endpoint episodes of suicidal and non-suicidal self-injuries; and change in depression severity. We expected that: (1) both treatments would lead to significant change after 14 months; (2) DBT would be more effective in reducing the number of episodes of suicidal and NSSI after 14 months; and (3) CM would be more effective in reducing depression severity after 14 months. These outcomes were chosen because DBT was specifically developed to treat behavioural dyscontrol, primarily self-harming and suicidal behaviour and we were interested in comparing CM on this outcome. In contrast, CM was developed to help patients develop a sense of self. There were no outcome measures with good 102 psychometric properties to evaluate this and hence, depression severity was used as a proxy. Recruitment started in January 2007, and the planned sample size of 162 was reached in April 2013. Post-treatment (14 month) data collection was completed in June 2014. Further follow-up assessments will occur over five years (see below), but will not be reported here (the time frame for data collection of follow-data is beyond the scope of the candidature). The study protocol was approved by the Hunter New England Human Research Ethics Committee (Reference Number: 06/12/13/5.11) and the University of Newcastle Ethics Committee (Reference Number H-2010-1146). See Appendix A for information sheet and for Appendix B for consent form. The trial is registered with the Australian New Zealand Clinical Trials Registry: ACTRN12612001187831 (registered 12 November 2012). 2.2.2 Study Setting This study was conducted at the Centre for Psychotherapy, a specialist outpatient service of the Hunter New England Mental Health Service (a public sector Mental Health Service), located in Newcastle, New South Wales, Australia. The Centre for Psychotherapy is a tertiary referral service that treats patients with BPD and/or Eating Disorders. The Centre accepts referrals from community mental health teams, general practitioners, and private therapists. The Centre treats patients falling within the geographical boundaries of the Hunter New England Health Service and comprises 10 full-time equivalent psychotherapists (including psychologists, psychiatrists, social workers, nurse therapists, and occupational therapists), and one dietitian. As part of the Australian public health care system, there are no treatment costs to patients attending the Centre. 103 2.2.3 Participants 2.2.3.1 Inclusion criteria. The inclusion criteria for this study were: a diagnosis of BPD according to Diagnostic and Statistical Manual of Mental Disorders 4th edition (DSM-IV) criteria (American Psychiatric Association, 1994). Diagnosis of BPD was made by a consultant psychiatrist using the Structured Clinical Interview for the DSM-IV (SCID-II; First, Spitzer, Gibbon, & Williams, 1995); a minimum of three episodes of suicidal and/or non-suicidal self-injurious behaviour in the previous 12 months assessed by a consultant psychiatrist during psychiatric interview (where suicidal and non-suicidal self-injurious behaviour is defined as acute, intentional self-injurious behaviour regardless of the intent to die, including both suicide attempts and other self-inflicted injuries such as cutting, deliberate self-poisoning or burning (Linehan et al., 2006a); and aged between 18 and 65 years. The inclusion criteria for the study were the same as the inclusion criteria for the clinical service and, as such, the study sample are representative of the patients generally seen at the service. As per standard practice for the service, no stipulation was made regarding medication status, and some patients were on medication and others were not. Their medication could be changed throughout the trial, as is standard for most BPD trials; and all prescribing occurred by practitioners independent to the clinical service. 2.2.3.2 Exclusion criteria. The exclusion criteria for this study were: disabling organic conditions; acute psychotic illness; antisocial behaviour that poses a significant threat to staff and fellow patients; developmental disability; living more than one hour’s drive from Newcastle; 104 inability to speak or read English; current substance dependence other than nicotine (as measured by SCID-I; eligible for entry once no longer meeting the criteria for dependence); and prior treatment with DBT or CM. 2.2.4 Content of the Interventions The template for intervention description and replication (TIDieR) checklist (Hoffman et al., 2014) was used as a guide in this section regarding what detail to include regarding the interventions. Patients were required to commit to at least 14 months of twice weekly psychotherapy in either program: twice weekly individual therapy sessions in the CM condition, and once weekly individual therapy session and a once weekly skill training group session in the DBT condition. After 14 months, the therapist and client made a mutual decision whether to continue treatment and contract for up to a further year of treatment. 2.2.4.1 Dialectical Behaviour Therapy (DBT). DBT is a manualised treatment (Linehan, 1993a; Linehan, 1993c) that combines treatment strategies from behavioural, cognitive, and supportive psychotherapies. It included weekly individual pre-treatment sessions for approximately four weeks (variable depending on when there was an available entry into the skills training group). Pre-treatment sessions included orientation to the treatment model, exploring goals, and eliciting a commitment to therapy. After the pre-treatment sessions, patients attended concurrent weekly individual and skills training group sessions for 12 months. The individual therapy sessions took approximately one hour per week and applied directive, problem-oriented techniques (including behavioural skill training, contingency management, and cognitive modification) alongside supportive 105 techniques. The skills training group met weekly for 2.5 hours and followed a psycho- educational format, teaching behavioural skills in three main areas in three modules: interpersonal effectiveness, distress tolerance, and emotion regulation skills. Preceding each module was a two-week focus on mindfulness, which was the core skill taught over the entire course. The skills group included the teaching and application of skills and their practice between sessions. In addition to attending individual therapy and skills training groups, participants had access to telephone coaching with their therapist during working hours. The DBT manual describes that phone coaching should be done in accordance with the therapist’s limits. For many therapists in the United States, that means being available individually seven days a week (Landes et al., 2017). At the service where the study was run, the arrangement that fitted the limits of the therapists in the service was to provide phone coaching during working hours, and outside of therapist working hours, participants had access to a telephone service staffed by a roster of six DBT therapists, which was available seven days per week from 8:30 a.m. to 10:00 p.m. This was the only modification to the standard DBT protocol (where individual therapists provide phone coaching outside of session time within their personal limits). DBT therapists attended a weekly consultation team meeting designed to provide support and assist therapists to be adherent to the treatment model. The 14-month schedule allowed for the pre-treatment phase in DBT to occur before participants commenced 48 sessions of skills training that occurred concurrently with the individual therapy. The 48 sessions usually took an entire year, taking into account that on occasional weeks the skills groups did not run (due to 106 public holidays, service closure for two weeks over Christmas or therapists being on leave). 2.2.4.2 Conversational Model (CM). Participants in the CM treatment arm attended twice weekly individual therapy for 14 months. The treatment model was developed by Meares (2004; Meares, 2005). At the time the trial commenced, an unpublished treatment manual was used to guide treatment. A published treatment manual is now available (Meares, 2012). The individual therapy sessions were approximately of one hour and were nondirective. The focus was on understanding the patient’s emotional experience and actively describing that experience back to the patient. The therapist actively looked for subtle signs of emotionally misunderstanding the patient, leading to mutual self- reflection and repair of the moment of disconnection in the therapeutic relationship. High value was placed on the patient’s real experience (as against socially acceptable experience) and the development of an authentic personal narrative. The patient was encouraged to find links between the maladaptive relationship patterns they had developed in their current social world and the relationship pattern they had with the therapist (and possibly, but not necessarily, the links with their childhood relationships). Fourteen months of CM therapy was selected as the treatment length to match that received in the DBT condition. Previous research on CM has involved treatment of one or two years’ duration. CM therapists attended weekly supervision either individually or in pairs. Patients could contact their therapist within working hours via phone for support, but not outside of working hours. Consideration was given to whether CM therapists would provide phone contact out of hours in order to match 107 that provided in the DBT arm. It was decided not to do so, given that this would be different from the original trials of CM and the aim was to evaluate the therapy as developed against an active treatment. In contrast, phone coaching outside of regular therapy appointments is a core part of DBT treatment model. 2.2.5 Therapists Thirty-two therapists, all with a minimum of two years clinical experience, delivered the treatment. They were clinicians who were employed to work in the Area Health Service and had not been hired specifically for this trial. Half of the therapists were employees of the Centre for Psychotherapy; some were employed in community or hospital mental health settings and served as visiting therapists to the Centre for Psychotherapy. Disciplines included psychologists, psychiatrists or psychiatric registrars, social workers, mental health nurses, and occupational therapists. Some therapists provided treatment in both treatment conditions (n=12), some provided treatment in DBT only (n=7), and some in CM only (n=13). Whether therapists elected to provide treatment in one or both conditions was the choice of the therapist; as such, this was not randomised. 2.2.6 Training Therapists providing treatment in the DBT arm of the trial were required to complete a minimum of four days of DBT training. All had completed additional two- day training sessions with the treatment developer, Marsha Linehan’s training company, Behavioral Tech (http://behavioraltech.org/index.cfm), and 14 of the therapists who delivered DBT completed 10-day intensive training in DBT in 2010 with Behavioral Tech. 108 Therapists providing treatment in the CM arm of the trial attended an introductory program; for the allied health and nursing staff, this involved a series of a minimum of 6 x two-hour seminars. For psychiatric trainees and psychiatrists, due to scheduling difficulties, the initial training was more individually tailored in meeting one on one with a staff specialist psychiatrist (2nd author) with extensive experience in the model to cover the seminar material. The majority of therapists in the CM arm of the trial had either completed or were completing a three-year part-time diploma specifically in CM with the treatment developer, Russell Meares’ training organisation, the Australian and New Zealand Association of Psychotherapy (ANZAP, http://www.anzapweb.com/). 2.2.7 Procedure 2.2.7.1 Recruitment, informed consent, and randomisation. Standard practice at the service was that upon receiving the referral, patients were allocated an appointment for an initial assessment with a psychologist, social worker, occupational therapist or mental health nurse therapist (all with a minimum of five years’ mental health experience) to determine whether they met eligibility criteria for the service and if there were any significant barriers that would interfere with their ability to make use of therapy. If at this assessment patients appeared to meet criteria for the service and were interested in pursuing treatment, they were assigned an appointment for a diagnostic interview with a consultant psychiatrist. Following diagnostic assessment, all patients meeting inclusion criteria for the trial were invited by a research assistant to participate in the study. Patients were provided with full explanation of the procedures and treatment conditions, including the stipulation that because it was a randomised trial, they would not be permitted to 109 change treatment allocation during the trial. If they were willing to participate, a signed informed consent was obtained. Participants received $20 per assessment to cover transport and related costs. Participating in research assessments was not obligatory to receiving treatment. Those patients who did not wish to consent to the study received the same treatment as those who consented to participate (either DBT or CM depending on preference and therapist availability). Consenting participants were randomly allocated to either DBT or CM. Figure 2.1 illustrates the overall flow of participants in the RCT. Details about the specific number of participants are included in Chapter 4. 110 Intake assessment Diagnostic assessment Exclusion Inclusion & Consent Baseline Assessment Randomisation DBT condition CM condition 7-month assessment (mid-treatment) 14-month assessment (post-treatment) Follow-up assessment 1 year after post-treatment assessment Follow-up assessment 2 years after post-treatment assessment Follow-up assessment 5 years after post-treatment assessment Figure 2.1. Flow of Participants in the RCT 111 A stratified centralised randomisation procedure was used to maximise the likelihood of comparable distributions across groups for gender and current treatment status for prescribed antidepressants. An independent research manager used a computerised formula with blocked randomisation (blocks of four). Participants were given a sealed, opaque envelope containing randomisation status at the end of their baseline assessment. Consequently, participants, therapists, and researchers were blind to treatment allocation prior to the participant opening the envelope. The research assistant completed all baseline assessments before the randomisation procedure. 2.2.8 Withdrawal Participants were informed that participation was voluntary and that they could withdraw from treatment or from the study at any time without negative consequences for their current or future care. For those who withdrew from treatment, we asked whether they would be willing to attend the remaining research appointments or at least provide minimal data (the primary outcome measures). 2.2.9 Assessments Data was collected at six time points: baseline, after seven months of treatment (mid-treatment), after 14 months of treatment (post-treatment), and at one-, two- and five-year follow-up post-treatment (1yr FU, 2yr FU, & 5yr FU). If patients went on to be part of the trial, part of the baseline information obtained during the diagnostic assessment with the psychiatrist was collected. As some clients continued with the therapy after 14 months and some stopped, at this point, whether 112 participants received intervention was no longer standardised, and the trial became a naturalistic one after post-treatment. Experienced research assistants with a minimum undergraduate degree in psychology conducted all research assessments, blind to treatment allocation status. Although the research assessments were conducted in the same building where therapists treat patients, the research assistant was scheduled to work at different times to when the skills training groups were run. Moreover, the research assistant regularly reminded both participants and therapists that they needed to be kept blind to treatment allocation; however, there was no formal measurement of the research assistant’s blindness. 2.2.9.1 Primary outcome measures. Table 2.1 summarises the data collection measures and schedule. There were two co-primary outcomes: (1) change from baseline to post-treatment assessment on suicide attempts and NSSI as assessed on the Suicide Attempt and Self Injury Count (SASI-Count; Linehan & Comtois, 1996; Linehan et al., 2006) and (2) scores on the Beck Depression Inventory II (BDI-II; Beck, Steer, & Brown, 1996). Given the nature of the target population and anticipated variations in treatment length and study retention, for planning purposes (e.g., resources, logistics, and statistical power calculations) the main outcome time point was post-treatment. 113 Table 2.1 Summary of Measures Mid- Post- 1yr 2yr 5yr MEASURE Baseline treat treat FU a FU a FU a ment ment Clinician-administered measures Primary outcome measure: SASI- X X X X X X Count (Suicide attempts & NSSI) BPDSI-IV (frequency and severity of X X X X X BPD symptoms) SCID-I for Depression, Anxiety, X X X X X Substance Abuse (Axis I disorders) SCID-II for all Axis II + screener (Axis X X X X X II disorders) Medication use X X X X X X Self-report patient measures Primary outcome measure: BDI-II X X X X X X (Depression Severity) IIP (Interpersonal Problems) X X X X X X KIMS (Mindfulness) X X X X X X DES (Dissociation) X X X X X X DERS (Emotion Regulation) X X X X X X SSI (Sense of Self) X X X X X X WAI – (Working Alliance Inventory) X X X Service utilisation X X X X X X Self-report therapist measures WAI – (Working Alliance Inventory – X X X therapist and client) a Not reported in the current analysis as data collection is ongoing for these time points The Suicide Attempt and Self-Injury Count (SASI-Count; Linehan & Comtois, 1996; Linehan et al., 2006) is a brief clinician-administered measure categorised into suicide attempts and non-suicidal acts within a specified period. The SASI-Count obtains information about the frequency of all self-injurious behaviours within a specified period by the method used, whether medical attention was required, and suicidal intent. This study utilised an adapted timeframe of behaviour occurring within the past seven months to match the assessment points. The SASI-Count is a briefer version of the Suicide Attempt Self Injury Interview (SASII; Linehan et al., 2006), which assesses self-injury in more detail. The SASII has demonstrated good reliability and validity. The SASI-Count was chosen over the SASII to reduce assessment burden. 114 The Beck Depression Inventory II (BDI-II; Beck et al., 1996) is a self-report instrument designed to assess and detect the severity of current (past two weeks) depressive symptoms in clinical, medical, and community settings. The BDI-II contains 21 descriptive statements regarding depressive symptoms frequently reported by individuals diagnosed with depression. Each of the items contains a four-point severity rating scale. Total scores range from 0 to 63, with scores between 14 and 19 considered “mild,” scores between 20 and 28 “moderate,” and scores above 28 regarded as “severe” levels of depressive symptoms (Crits-Christoph, Gibbons, Hamilton, Ring-Kurtz, & Gallop, 2011). It is a widely used, reliable measure of depressive symptoms. It has high internal consistency and correlates well with other self-report measures of depression and with clinician ratings of depression (r = .60-.90; see Beck, Steer & Carbin, 1988, for a review). 2.2.9.2 Secondary outcome measures. There were several secondary outcome measures: The Borderline Personality Disorder Severity Index (BPDSI-IV; Arntz et al., 2003) is a semi-structured interview based on DSM-IV criteria, which assesses the frequency and severity of various aspects of BPD. It has 70 items providing a total quantitative index ranging from 0 to 90 and the reference period is three months. A cut-off score of 15 identified BPD clients from non-patient controls with a specificity of 0.97 and sensitivity of 1.00, and scores less than 15 were considered indicative of recovery (Arntz et al., 2003). It has high inter-rater reliability and high internal consistency (Intra-class correlation = 0.97, Cronbach’s α = 0.93), as well as high concurrent and discriminant validity and sensitivity to change and is considered a suitable treatment outcome measure (Arntz et al., 2003; Spinhoven et al., 2007). 115 The Inventory of Interpersonal Problems (IIP; Horowitz et al., 1988) is a self- administered questionnaire for the assessment of subjectively experienced interpersonal difficulties. The IIP and its derivatives provide an overall quantitative index of interpersonal problems and scores on eight subscales: assured/dominant; arrogant/calculating; cold-hearted; aloof/introverted; unassured/submissive; unassuming/ingenuous; warm-agreeable; and gregarious/extroverted. Each subscale consists of eight items answered on five-point scales from 0 (Not at all) to 4 (Extremely). It has high internal consistency (Cronbach’s α = 0.96) and good convergent validity (Horowitz et al., 1988). The Kentucky Inventory of Mindfulness Skills (KIMS; Baer et al., 2004) is a 39- item self-report inventory with four subscales used for the assessment of mindfulness skills. Mindfulness is generally defined to include focusing one’s attention in a nonjudgmental way or accepting the experience occurring in the present moment (Baer et al., 2004). Participants indicate responses on a five-point Likert scale ranging from 1 Never or very rarely true to 5 Very often or always true and the instrument has good internal consistency (Cronbach’s α coefficients for “Observe,” “Describe,” “Act with Awareness” and “Accept without Judgement” subscales were .91, .84, .76, and .87, respectively) and adequate to good test-retest reliability (with corresponding correlations for these subscales being .65, .81, .86, and .83, respectively). It also demonstrates good content validity as well as good concurrent validity, correlating positively with other related measures. The Dissociative Experiences Scale (DES; Bernstein & Putman, 1986) is a 28-item self-report measure designed to quantify dissociation in normal and clinical 116 populations (van Ijzendoorn & Schuengal, 1996). It contains 28 items which describe experiences of disturbances in identity, memory, awareness, cognition and feelings of derealisation or depersonalisation (Bernstein & Putman, 1986). Respondents indicate the percentage of time they have experienced scenarios, with responses ranging from 0% to 100% (in increments of 10%). The mean of 28 items (the DES score) is an index of the number of different type of dissociative experiences and their frequency of occurrence (Bernstein & Putman, 1986). A meta-analysis conducted by van Ijzendoorn and Schuengal (1996) reported the DES to have strong internal reliability (mean Cronbach’s α = .93), very good convergent validity (combined effect size: d = 1.82; N = 5,916), and good predictive validity with Dissociative Identity Disorder (combined effect size d = 1.05; N = 1,705) and Post Traumatic Stress Disorder (combined effect size d = 0.75; N = 1,099). The Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004) is a 36-item self-report questionnaire designed to assess multiple aspects of emotion dysregulation during times of distress. The measure yields a total score as well as six subscale scores derived through factor analysis: (1) non-acceptance of negative emotions; (2) difficulties engaging in goal-directed behaviour; (3) impulse control difficulties; (4) lack of emotional awareness; (5) limited access to emotion regulation strategies; and (6) lack of emotional clarity. Participants indicate responses on a five- point Likert scale ranging from 1 Almost never to 5 Almost always. The DERS has been found to have high internal consistency (Cronbach’s α = .93), good re-test reliability, and adequate construct and predictive validity. The Sense of Self Inventory (SSI; Basten, 2008) is a 23-item self-report measure designed to identify several core components that reflect the subjective and 117 continuous experience of being an individual, authentic person who is in control of their own mental and physical activities. The subjective sense of self plays a significant role in both identity and psychopathology. It is the first known self-report assessment tool that aims to measure this concept. Participants indicate responses on a four-point Likert scale ranging from 1 Strongly Agree to 4 Strongly Disagree. Studies are currently underway that assess the scale’s reliability and validity. The Working Alliance Inventory (WAI; Horvath, Greenberg & Pinsoff, 1986) is one of the most commonly used and extensively validated measures of alliance. The WAI is a 36-item self-report measure with client, therapist, and observer versions of the questionnaire available and takes 10 minutes to complete. Both client and therapist versions were used in the current research. Baseline alliance was assessed after the sixth therapy session, allowing sufficient time for the alliance to begin to develop. Participants completed the client form (WAI-C) and therapists completed the therapist form (WAI-T). Both self-report questionnaires contain 36 items, rated on a seven-point Likert scale (1 = never and 7 = always), with 14 negatively worded and 22 positively worded items. All 36 items can be aggregated to create a total score, with higher scores indicating a stronger alliance. Twelve items tap into each of the three components of the therapeutic alliance as proposed by Bordin (1979): (a) tasks—the extent to which the therapist and client regard the tasks in therapy to be relevant (e.g., “I am clear on what my responsibilities are in therapy”); (b) goals—the extent to which the therapist and client agree on the goals of therapy (e.g., “The goals of these sessions are important to me”); and (c) bond—the positive personal attachment between client and therapist which includes trust, liking, acceptance, and confidence (e.g., “I am confident in ______’s ability to help me”). 118 The WAI is one of the most commonly used and well-established measures of alliance with excellent internal consistency (α = 0.93) for the client form and good internal consistency (α = 0.87) for the therapist form, and evidence to support convergent validity, concurrent validity and predictive validity (Horvath & Greenberg, 1989). 2.2.10 Treatment Adherence It is important to establish adherence to a treatment and to demonstrate that the intended treatment is actually delivered. Treatment adherence refers to the extent to which therapists use interventions consistent with the treatment manual for that particular treatment and avoid proscribed procedures (Waltz, Addis, Koerner, & Jacobson, 1993). Adherence was established through the use of treatment manuals and monitored through supervision. Therapists delivering DBT attended a weekly consultation team meeting, with approximately six members in each consultation team. Therapists who had only recently learned DBT also received individual supervision initially. For therapists providing CM, all therapists were involved in weekly supervision with an experienced CM supervisor either individually or in pairs. All individual therapy sessions were audiotaped with the patient’s permission for the purpose of supervision and to formally assess treatment adherence. Adherence scales developed for each treatment modality were used to evaluate adherence to the treatment manuals. Adherence raters were independent of the study and blind to treatment allocation. A randomly selected 5% sample of all sessions were rated for adherence. DBT sessions were rated by a coder trained in the use of the DBT Adherence Coding Manual (Linehan, 1993b) who also rated 30 CM sessions to 119 demonstrate treatment differentiation. In the absence of a suitable adherence scale for CM, one was developed for use for this study and individual treatment. A random 5% of CM sessions were rated by a coder trained in the use of the Newcastle Adherence Scale for Conversational Model, discussed in Chapter 3, who also rated 30 DBT sessions. Coders were not given information regarding the treatment model of the sessions they were rating. 2.2.11 Sample Size and Data Analysis 2.2.11.1 Sample size. The planned number of participants in the study was 162. This was based on the assumption that two-thirds would be retained at post-treatment (i.e., 54 participants per intervention condition). Family-wise Bonferroni-corrected statistical tests were employed throughout—namely, alpha divided by two for the primary outcome measures (e.g., 0.0250) and alpha divided by six for the secondary outcome measures (e.g., 0.0083). For example, regression-based analyses used to predict change scores on the primary outcomes at post-treatment included approximately six predictors (i.e., baseline scores for both outcome measures, age, gender, medication status at baseline, and group membership). It was expected that a sample of 162 participants should provide sufficient statistical power (80%) to detect modest population associations (e.g., simple or partial correlations) of 0.30 or higher for the primary outcome measures, using two-tailed statistical tests in the sample (or 0.34 or higher in the case of the secondary outcome measures). 2.2.11.2 Analysis. Analyses were planned to include a range of descriptive and analytical approaches, whereby all patients who attended a minimum of one treatment session 120 after the baseline assessment and were randomised were included in the analysis. Consistent with an intention to treat approach, all of the available data from the three time points were included in each analysis (not just complete pairs), using specific planned comparisons to examine differences between time points (and account for the repeated measures component of the variance). For consistency, the major outcome measures were expressed as change scores from baseline (T0), i.e., subsequent phase of either mid-therapy (T1) or post therapy (T2) minus baseline phase. Generalised Linear Modelling techniques (Generalised Estimating Equations, GEE) were used to examine differences between groups in the changes over time (with within-subject variation coded using subject IDs). For both primary and secondary outcomes, z scores were also calculated to show standardised change, expressed in relation to the grand SD of change. SPSS 22.0 software package for Window was used for statistical analyses. Chapter Six in this thesis focusses on minimally significant clinical differences to look at change at a more individual level. 2.3 Discussion The current study was the first “real-world” comparison of DBT and CM for the treatment of suicidal and non-suicidal self-injurious behaviour and depression severity among persons with BPD. It aimed to extend our knowledge about how well DBT translates to public mental health treatment settings, with therapists in this trial being those who were routinely employed by the health service. Although CM has been the focus of a number of studies, none have involved randomisation. It was decided not to conduct a randomised study comparing CM against treatment as usual as the existing 121 data appeared sufficiently promising to believe that this was an effective treatment. Thus the appropriate next step for investigation was to compare CM against an established active treatment. 2.3.1 Strengths and Limitations There were several strengths to the current study. The service had already been the setting of an RCT comparing six months of DBT with TAU plus a waiting list control (Carter et al., 2010). As such, the feasibility of running an RCT at this service had already been established. The sample size for the current trial was sufficiently large to recognise a difference in the primary outcomes; much larger than most previously published RCTs in the field of BPD. Outcomes were assessed using well- validated measures commonly used in other BPD trials and participants were and will be followed up at one-, two-, and five-year time periods. Stoffers et al. (2012) in the most recent Cochrane review of psychological therapies for BPD suggest “there is an urgent need for independent research endeavours” (p. 77), and the current trial goes some way to providing such evidence, given that none of the authors are treatment developers nor have they worked clinically alongside the treatment developers. Several elements in the trial were designed to reduce potential bias. The stratified randomisation based on gender and antidepressant medication allowed us to reduce the bias on outcome that may occur if imbalance at baseline occurred in these two areas without randomisation. Blindness of raters for both outcome variables and adherence ratings further reduces potential bias of outcomes. Adherence coding for both treatment arms commenced after the trial was initiated. It was decided that it was important that therapists did not receive the results of the adherence coding during the trial, as it is unrealistic that such feedback 122 would generally occur in a real world setting, and feedback provided mid-way through the trial would have introduced a confounding variable in terms of outcome. The combination of these factors would undermine our capacity to obtain an accurate picture of the effectiveness of the interventions in a routine clinical practice setting. The benefit of this situation is that it will allow us to explore questions that have not been addressed to date in the BPD literature, for example, whether the level of therapist adherence to a treatment model is related to outcome. Stoffers et al. (2012) criticised the majority of trials of psychological therapies for BPD on the basis of the amount of professional contact, in that control group participants did not receive comparable amounts of professional attention. In the current study, participants in the CM condition attended twice weekly individual therapy, while those in the DBT condition attended once weekly individual therapy and once weekly skills training. The amount of attention received from professionals was comparable, even though those in the CM group received two hours of face-to-face individual therapy and those in DBT received one hour of face-to-face individual therapy and 2.5 hours of skills training (with two therapists and usually eight people in a group). Ideally, a third arm, such as minimal treatment control group, would have been included. However, it was deemed unethical for persons at known risk of suicide to be denied an evidence-based psychotherapeutic treatment when one was available. The main limitation of the current RCT was likely to be participant retention. Unlike most efficacy trials, there was little incentive for participants to attend assessments as they received treatment regardless of their participation in the study. There were a number of weaknesses that are consistent with the study being completed in a real world setting. Clinical supervision addressed practical adherence 123 issues on an ongoing basis, including listening to tapes of sessions and feedback being given by peers and expert supervisors. However, as formal adherence ratings occurred retrospectively, it was only on completion of the trial that we knew to what extent the therapists were adherent to the model. If they were less than adherent, there would be potential implications for the generalisability of the results, since the implication is that therapists had not delivered the treatment in its entirety. In most clinical trials, therapists only deliver one of the interventions being evaluated. In the current study, some therapists delivered only one of the models and some delivered both. This is a less pure arrangement, and it is unclear to what extent those therapists who deliver both models can do so adherently. This is an empirical question, and adherence ratings will enable us to explore whether therapists delivering two models are less adherent than therapists only delivering one model. Although the Centre where the study was undertaken sits within public sector mental health services, it is a specialist service for BPD and Eating Disorders. As such, while the therapists were not specifically hired for the study and were more representative of therapists found in real world settings, they did have expertise in BPD beyond that which is likely to be found in a general community service; this may limit the generalisability of the findings to general treatment settings. These issues will be expanded upon in Chapter 7. 124 Chapter 3 Newcastle Adherence Scale for Conversational Model (NASCOM): Development and Utility of an Adherence Measure for Conversational Model Psychotherapy Chapter 2 outlined the methodology of the RCT including discussion of the specific treatments. In conducting a clinical trial, therapist adherence to the treatment manual needs to be assessed. While there was an existing adherence measure for DBT, no such measure existed for CM. Hence, it was necessary to develop a measure that could be used to assess adherence in the trial. Chapter 3 outlines the development of this measure and its utility for assessing adherence. The NASCOM was originally developed by Dr Bernard Goldman, under the supervision of Dr Carla Walton and Professor Mike Startup. Some content from Chapter 3 was included in the thesis submitted by Dr Goldman in partial fulfilment of the requirements for the degree of Doctor of Clinical Psychology, University of Newcastle. It is included here with permission of Dr Goldman and Professor Startup (now deceased). 2 January 2018 Dr Bernard Goldman Date 125 It is important when evaluating psychotherapies to know that the core components of the therapy were delivered; otherwise, it is not possible to make any definitive conclusions about the outcomes of the study. Treatment manuals are now regarded as necessary in psychotherapy outcome studies in order to ensure that the treatment delivered is standardised. There is also an expectation that in addition to using treatment manuals, researchers document the extent to which the treatment manual was followed (Waltz et al., 1993). At a basic level, adherence is defined in a checklist format noting either the occurrence or non-occurrence of a specific technique. The next level is to measure the proportion of particular components against the overall content of the therapy session. Subsequently, the therapy should be shown to be distinct from other therapies. In contrast to adherence, competence is typically measured by “experts” who rate the level of skills exhibited by the therapist in delivering the treatment technique (Waltz et al., 1993). Perepletchikova, Treat, and Kazdin (2007) found that adherence testing was implemented significantly more in studies of skill-building treatments such as cognitive behaviour therapy than in non-skill-building treatments such as psychodynamic therapy. They note that skill-building approaches generally focus on symptom reduction by means of techniques which form part of an agreed plan of interventions. Perepletchikova et al. (2007) argue that in contrast to skill-building therapies, psychodynamic treatments focus on insight and usually involve exploration of underlying themes and patterns in patients’ lives and consequently are less easy to rate than skill-building therapies. As adherence testing is one of the building blocks to evaluating evidence-based treatments, there is a need to undertake adherence testing for psychodynamic therapies. With funding agencies and insurance companies 126 requiring evidence-based treatments, psychodynamic therapies risk becoming obsolete if they are not evidence-based (Chambless & Ollendick, 2001; Guthrie, 1999). There are a number of psychodynamic treatment modalities that have developed adherence scales for use in studies evaluating those therapies, although they do not appear to have published findings regarding the psychometrics of the scales (e.g., TFP; Yeomans, Hull, Delaney, & Clarkin, 2004). However, there are adherence scales for psychodynamically oriented therapies which have been developed, tested, and described in the published literature, for example, Accelerated- Experiential Dynamic Psychotherapy Fidelity Scale for Accelerated-Experiential Dynamic Psychotherapy (AEDP-FS; Faerstein & Levenson, 2016) and Vanderbilt Therapeutic Strategies Scale (VTTS; Butler, Henry, & Strupp, 1995). There are three published adherence scales developed for psychotherapies addressing BPD. Karterud et al. (2013) undertook an extensive process to develop the Mentalization-Based Treatment Adherence and Competence Scale (MBT-ACS), which as the name suggests measures both adherence and competence. They found the MBT-ACS to be a reliable instrument for rating adherence and competence, with higher reliabilities among seven raters (.84/.88, respectively) than when reduced to using two raters (.60/.68, respectively). They also found considerable variability in the reliabilities of individual items. Kolla et al. (2009) developed the General Psychiatric Management Adherence Scale for General Psychiatric Management (GPM), a psychodynamically informed therapy. The General Psychiatric Management Adherence Scale was used in a randomised trial in which GPM was compared with DBT (McMain et al., 2009). The Scale was tested by measuring therapist and client agreement for interventions undertaken in therapy by way of questionnaires administered during the 127 course of the therapy. It was not used to test for discrimination between GPM and DBT. An adherence instrument has also been developed for Dynamic Deconstructive Therapy (DDT) for treatment-resistant clients with BPD and has been found to have excellent inter-rater reliability (0.91) within a sample of 10 participants (Goldman & Gregory, 2009). While DDT is regarded as psychodynamically oriented, the DDT adherence scale appears to focus on the directive components of that therapy. The Sheffield Psychotherapy Rating Scale (SPRS; Shapiro & Startup, 1990) was developed to measure adherence to Exploratory Therapy. Shapiro and Startup (1992) found satisfactory inter-rater reliability for most items in the SPRS. Startup and Shapiro (1993) also found that the ratings contributed significantly to discrimination between the Exploratory and Cognitive Behaviour therapies used in that study (see Shapiro & Startup, 1992). The development and testing of any adherence instrument requires an understanding of the therapy being rated. CM is a psychodynamically oriented therapy originally developed by Robert Hobson (1985) that focuses on the therapist-client relationship as the avenue for exploring and resolving interpersonal difficulties. The therapy’s name reflects the process of interpersonal learning occurring through a focused “conversation” in the “here and now” about the patient’s feelings and interpersonal problems and how they present themselves in the therapeutic relationship (Margison & Shapiro, 1986). It was originally used in the treatment of people with affective disorders, where problems in relationships were considered to play a part in causing or maintaining the disorder. It is also known as “Exploratory Therapy” (Startup & Shapiro, 1993) and a briefer version, as “Psychodynamic Interpersonal Therapy” (PIT; Shapiro et al., 1994). 128 For Hobson, one of the key issues in therapy was to address the client’s fear of aloneness or abandonment. He found that when the client can experience being alone, he or she is able to experience togetherness which does not involve fusion with another person. Hobson suggested that intimacy fails when clients have felt a sense of loss or abandonment in their relationships. One of the key goals of CM is for the client to learn to be “alone and together” especially in the therapeutic relationship (see also (Gunderson, 1996). Such a goal is appropriate for the client with BPD who experiences unstable relationships and feelings of abandonment. Hobson (1985) also posited that disturbance in a client’s relationships may arise out of lack of opportunity in the client’s environment to learn a language which allows feelings to be expressed or understood. CM aims to assist the client to find a language which allows them to converse with the therapist about their feelings in a way that they can use in other significant relationships. The therapy aims to help the client to change relational behaviours that have impaired them in the past or may cause or maintain various distressing symptoms (for example, avoidance of intimacy). Hobson promoted the exploration of the client’s feelings in an active experiential way in therapy. He emphasised the importance of the therapist observing the client’s non- verbal language as another means of exploring the client’s feelings. He regarded the therapist’s task as being to encourage the client to stay with his or her feelings. CM, as developed by Hobson (1985), involves an integration of experiential/humanist, interpersonal, and psychodynamic interventions (see Shapiro & Startup, 1992). Experiential/humanist interventions cluster around the therapist emotionally attuning to the client in a tentative and collaborative way. Hobson encouraged a collaborative approach by way of mutual feeling language (the therapist 129 using “we” instead of “you”), the making of tentative statements, and avoidance of therapist questioning. The key intervention for change in CM is the therapist staying in the “here and now” of therapy by focusing on client feelings (stated and unstated). Interpersonal interventions include exploration of patterns in previous relationships. Psychodynamic interventions include the provision of explanatory statements which provide possible reasons or causes for clients’ behaviours such as links with the client’s childhood. A list of CM strategies rated in this study is found in Table 3.1. 130 Table 3.1 Newcastle Adherence Scale for the Conversational Model (NASCOM) (Unless otherwise stated, every item begins with the phrase, “Did the therapist...”) Item Item description 1 Receptive listening ... allow silence to continue (or use minimal encouragement as a means of encouraging the client to talk? 2 Tentative style ... express his/her views as tentative statements, open to correction, and inviting elaboration and feedback? 3 Language of mutuality ... use the language of shared endeavour (“I” and “we” and the passive tense)? 4 Patterns in relationships ... draw parallels or point out patterns in two or more of the client’s relationships for the purpose of helping the client understand how she/he functions in interpersonal relationships? 5 Therapeutic relationship ... addressing client’s feelings about the therapeutic relationship? 6 Emotional attunement ... attune to the emotional cues or other words offered by the client by using either of the following micro-skills: coupling, amplification, or representation? 7 Awareness of feelings ... encourage client to consider unspoken feelings of which the client may be unaware or avoiding? 8 Avoidance of affect ... assist the client address any avoidance experienced by the client? 9 Acceptance of affect ... encourage the client to accept feelings of which the client is aware but which are painful or uncomfortable? 10 Explanatory statements ... introduce explanatory statements which offer possible reasons for the client’s behaviour and experiences? 11 Metaphor ... encourage and elaborate the client’s use of metaphor? 12 Personal disclosure ... respond to client’s personal questions ...in a way that advanced a shared understanding of events and processes in therapy? 13 Limitations ... promote the client’s exploration of feelings concerning the limits to therapy, and boundary, loss and internalisation issues related to termination? 14 Disjunctions ... address any disjunctions in therapy (i.e., addressing occasions when client ceases to respond in therapy often due to the therapist failing to attune to the client)? 15 Frame changes ... address frame changes (i.e., addressing occasions when the client’s behaviour changes in therapy e.g., attending sessions late)? 16 Warmth ... convey warmth? 17 Rapport How much rapport was there between therapist and client? 18 Agenda Setting To what extent did the therapist set out an agenda for the session? 19 Directiveness How much did the therapist direct or guide the session in an explicit way? 20 Providing reassurance ... provide reassurance to the client? 21 Advice giving ... provide non-psychological advice and undertake problem solving? 22 Psychological techniques ... offer psychological techniques to assist the client or suggest various types of practice of techniques between sessions? 23 Psycho-education ... provide psychoeducation around various issues affecting the client? 24 Information gathering ... gather information by way of questioning which was not for the purposes of clarification? 25 Homework assigned ... develop one or more specific assignments for the client to engage in between sessions? 131 While Hobson’s and Meares’ early focus was on interpersonal disturbance and its impact on mental well-being generally, Meares (2005) adapted CM to those persons with a diagnosis of BPD. Meares was interested in the internal experience of clients with BPD and the damage to their sense of “self,” which he describes as an internal flow of feelings, thoughts, and meanings. A healthy self enables the person to experience intimacy in relationships and to retain the capacity to experience aloneness. This sense of self is compromised or stunted in BPD, particularly at times of high emotional arousal. The task of the therapist in CM is to attune to the emotional experience of the client, so as to develop a shared understanding of the client’s internal feelings, thoughts, and sensations. This enables the client to make sense of their behaviour in the world, particularly in close relationships, both past and present. The therapist encourages the client to express and discuss experiences, feelings, and behaviours both within the therapeutic relationship and in the outside world, as the client with BPD often does not have the words to describe their feelings and internal experiences. At times in the therapy, this collaborative understanding will break down, leading to a rupture within the therapeutic relationship, called a “disjunction” in CM. Understanding and resolving these disjunctions collaboratively allows the client to develop a capacity to feel more stable within, and to resolve their difficulties in other relationships. Significant reductions in violent behaviours, use of drugs, medical visits and hospital admissions, self-harm episodes, and time away from employment occurred when CM was delivered twice weekly over one year (Stevenson & Meares, 1992). The gains made in therapy were found to continue in a five-year follow-up (Stevenson et al., 2005). A replication of the study by Stevenson and Meares (1992) also found 132 similar gains (Korner et al., 2006). More detail about these studies is provided in Chapter 1. However, there was no adherence testing in the above-mentioned studies. CM has also since been expanded to the treatment of other complex disorders such as treatment resistant depression (Stevenson, Haliburn, & Halovic, 2015). As discussed, Exploratory Therapy and Psychodynamic Interpersonal Therapy are related therapies to CM that share its early origins. These therapies used the Sheffield Psychotherapy Rating Scale (SPRS; Shapiro & Startup, 1990), which was co- developed by Mike Startup, an associate investigator in the development of the NASCOM. With permission from the developers, some of the items in Table 3.1 reflect the adherence items for Exploratory Therapy in the SPRS. Perepletchikova et al. (2007) raised but did not answer the question about whether integrity measures need to be devised for each treatment. However, if adherence testing is to promote the internal and external validity of the treatment under study, adherence needs to relate to the treatment as conducted. It is arguable that CM, as developed by Meares (2005) for the treatment of people diagnosed with BPD, is sufficiently distinct from the Exploratory Therapy used in the treatment of employed persons with depression by Shapiro and Startup (1992) that a new adherence scale was required. Meares (2004, 2005) adapted the principles of CM as developed by Hobson for people with BPD, who have been found to be not as high functioning as people suffering with a primary diagnosis of depression (see Gunderson et al., 2011). As a result, Meares (2005) developed a long-term therapy of at least 12 months’ duration where the therapists’ tasks were to attend closely to the client’s emotions from moment to moment with the aim of building a therapeutic relationship and facilitating the development of the inner world of the client. Meares emphasises 133 the importance of addressing any issues arising out of the therapeutic relationship in CM, as he had found people with BPD generally have experienced a long history of unstable relationships and on that basis, there is an expectation that their experiences will impact negatively on the therapeutic relationship (Meares, 2005). Meares published his treatment manual for CM in 2012. The Conversational Model Clinician’s Manual for the Treatment of BPD (Meares, 2012) includes the basis of an adherence scale; however, there are no published reports about the psychometrics of this scale. The Newcastle Adherence Scale for the Conversation Model (NASCOM) and its corresponding manual were created specifically for rating adherence to CM. Such a scale was needed to assess treatment adherence within a Randomised Clinical Trial (RCT) conducted at the Centre for Psychotherapy, a public sector specialist service for the treatment of BPD, as discussed in Chapter 2. That RCT compared CM and DBT in the treatment of adults with BPD. DBT is a cognitive-behavioural based therapy developed specifically for the treatment of BPD (Linehan, 1993a). The current study has three aims: (1) to describe the development of the NASCOM as an adherence scale for CM; (2) to examine inter-rater agreement; and (3) to examine the capacity of the NASCOM to discriminate between CM and a cognitive- behavioural therapy. On the basis of the findings in Shapiro and Startup (1992), it is hypothesised that the NASCOM would have reliable inter-rater agreement. Further, on the basis of the finding in Startup and Shapiro (1993), it is hypothesised that the NASCOM would be able to distinguish CM from DBT in the current RCT. 134 3.1 Method 3.1.1 Participants As the study involved evaluating an adherence tool, participants included both patients who were part of the RCT and the therapists delivering the intervention. 3.1.1.1 Patients. Participants were part of an outpatient RCT for the treatment of BPD. As the current study focuses on the development of the adherence scale, the clinical trial is only briefly described here to provide context for this study; see Chapter Two for more detailed descriptions of the participants and the study. In brief, participants had a diagnosis of BPD, were between 18 and 65 years of age, and had reported deliberate self-injury on at least three occasions in the previous 12 months before intake interview. If they provided consent, they were randomised to either CM or DBT and received 14 months of treatment: twice-weekly individual therapy in the CM arm or once weekly individual therapy and once weekly skills training in the DBT arm. A sample of 20 participants from the RCT were included in the testing of the NASCOM. This was a subset of the total sample of 162 participants included in the RCT. Diagnosis was made using the Structured Clinical Interview for DSM-IV for Axis II Disorders (First, Gibbon, Spitzer, Williams, & Benjamin, 1997) as well as a diagnostic interview conducted by a consultant psychiatrist. There were seven male and 13 female participants. The mean age was 27.5 years (SD = 6.7 years). Most (18) of the participants described themselves as white Caucasian, with the other two participants of Indigenous Australian or African descent. 135 3.1.1.2 Therapists. Sessions from 14 therapists involved in the RCT were included in this study. In total, there were 32 therapists in the RCT, of whom seven delivered DBT only, 13 delivered CM only, and nine conducted both therapies. A priority was given to sampling all therapists practising in both modalities who were working with a substantial number of clients in the study, in order for the sessions to be most representative. As a result, six therapists were selected from the group of therapists who delivered therapy in both modalities. The mean years of experience in practising CM and DBT for this group were 5.6 years and 7.8 years respectively. Further, an equal number of therapists practising CM only and DBT only were selected (i.e., four therapists from each modality) and emphasis was placed on those who were seeing more than one client in the study. The mean years of experience for therapists practising CM only and DBT only were eight years and 12.5 years respectively. All therapists involved in the study were supervised on a weekly basis. All CM therapists were supervised by a faculty member of the Australian New Zealand Association of Psychotherapy (ANZAP), the principal training institute for CM in Australia. All CM therapists had a minimum six sessions of training. Seven of the therapists had completed or were in the process of completing the Diploma of Adult Psychotherapy, a three-year part-time diploma that involves intensive training in CM. 3.1.1.3 Raters. The raters (Mike Startup & Bernard Goldman) were both clinical psychologists. While B.G. had six one-hour sessions of training in CM, neither rater had practised CM with clients with BPD. While B.G. had observed group sessions of DBT, neither rater had been trained in DBT. 136 3.1.2 NASCOM Scale Development The NASCOM consists of 25 items and was designed to be rated by an observer. It was created by B.G. & M.S. after an extensive and iterative consultation process with the thesis author and another investigator (Nick Bendit), both of whom are experienced practitioners in CM. As such, we expected that the scale had face validity, since all items were included with the approval of expert clinicians in CM. The NASCOM is included in Appendix 3. A summary of the items in the NASCOM is found in Table 3.1. The first 15 items in the NASCOM involve CM interventions. The next two items are general facilitative conditions in psychotherapy, namely, use of warmth and rapport. The remaining eight items are proscribed interventions which are unacceptable in CM. However, all the proscribed items either form part of cognitive behavioural therapies such as DBT or are acceptable to such therapies. The NASCOM was largely modelled on the SPRS. Similar to the SPRS, each item in the NASCOM has a seven-point Likert rating scale. However, the range in the NASCOM is from 0 (representing non-occurrence of a particular intervention in a session) to 6 (representing use of an intervention in 31% or more of the session). Each point on the rating scale generally represents a 5% increment in the session time (e.g., 1: 0 to 5%; 2: 5 to 10%; to 6: 31% or more). A rater manual defines each item. The NASCOM manual (see Appendix C) attempted to be as consistent as possible with the format of the SPRS manual in the following ways: (1) the exact wording of the NASCOM was reflected in the manual; (2) an elaboration of the item’s purpose and any definitions that seemed appropriate in relation to the item; (3) examples of adherent and non-adherent behaviour; and (4) advice on how to distinguish items in the NASCOM. Items were written or modified so as to focus exclusively on the therapist 137 behaviour (in contrast to the patient’s behaviour). Hence, each item commences with “did the therapist…” Further, to increase reliability, items were worded in order to focus on observable therapist behaviours and to be as specific and concrete as possible. While the NASCOM used the SPRS as its model, about 70% of the items in the SPRS were removed, as it was not expected that they would be used in CM. Of the 59 items in the SPRS, 30 related to a Cognitive Therapy Scale and were excluded. Eight items from the SPRS come from a Facilitative Conditions Scale, of which two items were included in the NASCOM. Of the 16 items reflecting Exploratory Therapy in the SPRS, 12 were adapted to the NASCOM. While some items in the NASCOM were similar to the items in the SPRS, the examples of adherent and non-adherent behaviour differed, to reflect CM practice. A detailed table of which items were included from the SPRS and which were excluded is found in the NASCOM manual. Pilot coding of the NASCOM used 12 de-identified sessions from an archived bank of CM sessions with patients with BPD and was undertaken by B.G. and M.S. The pilot phase allowed for an iterative process of amendments to the NASCOM and its manual so as to minimise any ambiguities. It was intended that subsequent to the 12 sessions coded to finalise the manual, ratings would be conducted for 60 sessions, representing 10 DBT clients and 10 CM clients, with three sessions for each client selected from different periods of therapy. Over the 14 months of therapy, the three time periods identified were “early” (first four months), “mid” (second four months) and “late” (final five months of therapy). While 60 sessions only amounted to a small percentage of sessions involved in the RCT, the purpose of collecting 60 sessions was to develop and validate the NASCOM rather than to test adherence to CM in the RCT. 138 It was anticipated that the greatest chance of non-adherence to therapy would take place where therapists were practising in both modalities, albeit with different patients. On that basis, as can be seen in Table 3.2, sessions of six therapists using both modalities were randomly sampled making a total of 36 sessions (i.e., six therapists (6) x treatment in both therapies (2) x three sessions per client (3)). Sessions from the four therapists practising CM only and DBT only were sampled making a total of 24 sessions (i.e., four therapists in CM plus four therapists in DBT (8) x three sessions per client (3)). One client’s third phase of therapy became unavailable as the client did not commence the last five months of therapy; thus only 59 sessions were rated. 139 Table 3.2 Breakdown of Sessions Coded According to Therapy Model, Phase, and Therapists Therapist CM Sessions DBT sessions Total Phase of therapy Phase of therapy Early Mid Late Early Mid Late Both treatment models: Therapist 1 Client 1 X X X Client 2 X X X 6 Therapist 2 Client 3 X X X Client 4 X X X 6 Therapist 3 Client 5 X X X Client 6 X X X 6 Therapist 4 Client 7 X X X Client 8 X X X 6 Therapist 5 Client 9 X X X Client 10 X X X 6 Therapist 6 Client 11 X X X Client 12 X X X 6 CM only: Therapist 7 Client 13 X X X 3 Therapist 8 Client 14 X X X 3 Therapist 9 Client 15 X X X 3 Therapist 10 Client 16 X X X 3 DBT only: Therapist 11 Client 17 X X X 3 Therapist 12 Client 18 X X X 3 Therapist 13 Client 19 X X X 3 Therapist 14 Client 20 X X N/A 2 Total 10 10 10 10 10 9 59 sessions The first four weeks of therapy sessions were excluded from ratings as the researchers wanted to ensure that ratings were based on sessions that truly represented the therapies and not assessment or orientation to the therapy. Three sessions were randomly selected for each of the clients, with one session from the early (first four months), mid (second four months) and late (final five months of therapy) phases. Therapists used digital recorders for taping complete sessions across the entire course of therapy (with client consent elicited when they enrolled in the study). Both therapists and raters were blind with regard to what sessions in the entire treatment would be chosen for later ratings. Sessions were chosen at random within the session blocks described above. 140 3.1.3 Procedure A total of 20 sessions were rated by both M.S. and B.G. using the NASCOM to test the internal validity of the NASCOM. This number of sessions should be sufficient to adequately test the validity of an adherence testing instrument (Shrout & Fleiss, 1979). One session was randomly selected from each therapist-client dyad across the different phases of therapy. Hence, 10 sessions involved CM and 10 involved DBT. The type of therapy was not disclosed to the raters until all ratings had been completed. The sessions rated were audiotaped as well as fully transcribed by B.G. and made available to M.S. Raters based their ratings on the audiotapes and the transcriptions. The use of transcriptions assisted in determining the percentages of each strategy in a session. After the 20 joint ratings were completed, B.G. rated 20 sessions alone and then undertook a joint rating with M.S. Following a subsequent 20 sessions, a further joint rating session took place. This was undertaken to check for possible “rater drift,” and no drift was identified. All 59 sessions were also rated on the DBT adherence scale by a coder trained to a reliability standard in the use of that scale. 3.1.4 Statistical Analyses The data was analysed initially using SPSS (Version 19) for Windows (SPSS Inc., Chicago, Il, USA) and subsequently using SPSS (Version 23). Inter-rater reliabilities were estimated as intraclass correlation coefficients (ICC) using methods proposed by Shrout and Fleiss (1979). For the purposes of deriving a global CM adherence score (referred to subsequently as the “Total score”) each of the 25 items was classified as falling into one of the following categories: (1) essential to CM and expected to occur in high 141 frequency; (2) desirable but not essential; or (3) proscribed in CM (i.e., largely prohibited strategies, such as cognitive-behavioural techniques). The following six items were classified as essential to CM and expected to occur in high frequency: Receptive listening, Tentative style, Language of mutuality, Emotional attunement, Warmth and Rapport. The following eight items were considered proscribed in CM: Agenda setting, Directiveness, Providing reassurance, Advice giving, Psychological techniques, Psycho-education, Information gathering, and Homework assigned. The remaining 11 items were considered desirable but not essential. Items designed as essential to CM and expected to occur in high frequency were given a score of 1 if they occurred in 21% or more of the session; items designated as desirable but not essential were given a score of 1 if they occurred in 6% or more; and proscribed items were given a score of -1 if they occurred in 6% or more of the session. These item scores were summed to generate a Total CM adherence score, which could potentially have ranged from -8 to 17. In addition, CM subscale scores were calculated in a consistent manner with those utilised in the SPRS (by averaging the original 0 to 6 ratings), including: Facilitative conditions (3 items), Shared understanding (4 items), Focus on feelings (5 items), Gaining insight (3 items), Therapeutic ruptures (2 items), and CBT items (8 items). Discriminant analyses were performed by comparing the scores on the NASCOM and the scores on the DBT Adherence Scale for the selected CM versus DBT modality therapy sessions. 142 3.2 Results 3.2.1 Inter-rater Agreement Item level inter-rater agreements for adherence to the CM items are presented in Table 3.3, across both CM and DBT modality sessions in the first data column and based solely on CM modality sessions in the second data column. 143 Table 3.3 Psychometric Properties of Items in the NASCOM Reliability Presence of item Item Abbreviated title Across Within CM Proportion Proportion modalitiesa modality of CM of DBT number sessionsb sessionsc sessionsd 1 Receptive listening .90 .86 93.4% 91.7% 2 Use of tentative style .82 .78 93.4% 91.7% 3 Use of language of mutuality .84 .85 100% 91.7% 4 Identifying patterns in .37 ns .52 ns 50% 25% 144 relationships 5 Therapeutic relationship .83 .90 50% 25% 6 Emotional attunement .91 .80 100% 100% 7 Awareness of feelings .58 .56 ns 86.8% 75% 8 Avoidance of affect .66 1.00 6.6% 8.4% 9 Acceptance of affect .39 ns Ns 33.4% 16.7% 10 Explanatory statements .73 .68 86.8% 75% 11 Use of metaphor .88 .93 13.2% 16.2% 12 Personal disclosure .75 -.82 ns 25% 25% 13 Exploration of limitations .88 Ns 6.4% 8.4% 14 Disjunctions .62 .78 13.2% 16.7% 15 Frame changes .91 .90 25% 16.7% 16 Warmth .55 .21 ns 100% 100% 17 Rapport .43 ns .32 ns 100% 100% 18 Setting agenda .81 .00 ns 0% 100% 19 Directiveness .82 .1 ns 33.4% 100% 20 Providing assurance .87 .2ns 13.2% 91.7% 21 Providing advice .90 .78 13.2% 91.7% 22 Psych techniques .95 .00ns 0% 100% 23 Psychoeducation .95 .78 13.2% 100% 24 Providing information .92 .80 50% 100% 25 Setting homework .75 .00ns 0% 91.7% aInter-rater reliability based on sessions of both CM and DBT. bInter-rater reliability for CM sessions alone. c Percentage of CM sessions in which the component was rated as present by both raters. . d Percentage of DBT sessions in which the component was rated as present by both raters. ns Reliability not significant from zero. 145 In summary, there was good inter-rater reliability against a number of key items used in the CM, as well as for proscribed items in CM. As would be expected, reliabilities for the items were generally higher in most cases when both treatments were included in the analysis, reflecting the differences in the delivery of therapy between the two modalities. The third data column in Table 3.3 identifies the percentage of sessions jointly rated in which the item was present (at all) in the CM. The fourth column outlines the percentage of sessions in which the item was present (at all) in DBT. When testing for inter-rater agreement across the CM and DBT modalities, 14 of the first 17 (CM relevant) items in Table 3.3 were found to have significant levels of inter-rater reliability. When testing within CM modality, 10 items received significant levels of inter-rater reliability. It is of note that the lack of reliability for non-significant items was largely due to the interventions being rated either as “not present” (i.e., scoring 0) or almost always present (e.g., Warmth and Rapport). With the CM items which did not reach a significant inter-rater agreement across modalities, rater disagreement was found in only one case greater than two points. Rater disagreement was generally no greater than one point difference for all items. Apart from generic, facilitative conditions (Warmth and Rapport), only six CM items were jointly rated in 85% or more of the CM modality sessions. They were: Receptive listening, Tentativeness, Mutuality, Emotional attunement, Facilitating awareness of emotions, and the Use of explanatory statements. “Exploration of patterns in relationships” and “Addressing issues in the therapeutic relationship” were jointly rated in 50% of the CM sessions. “Avoidance of affect,” “Emotional acceptance,” and “Use of metaphor” were jointly rated less frequently. 146 Ratings for all proscribed items were found to have significant inter-rater reliabilities when using both modalities. Within the CM modality alone, three proscribed items were not present (i.e., Setting agendas, Psychological techniques, and Setting homework), which accounts for the non-significant inter-rater reliabilities. Further, as the third and fourth data columns in Table 3.3 show, in the CM modality sessions proscribed items were rarely used; however, they were present in almost all DBT modality sessions (as would be expected given that these are active components of that treatment). 3.2.2 Discriminant Analysis The validity of the NASCOM was tested chiefly by examining its capacity to discriminate between the two treatment modalities. Means on each of the adherence subscales and total scores are shown in Table 3.4, ordered from highest to lowest for DBT sessions and from lowest to highest for CM sessions (on the relevant adherence scale). A one-way ANOVA revealed that the DBT and CM sessions were significantly different on the majority of DBT subscales and on the total DBT score. For the seven DBT subscales significantly different at P<.01 or better, the variance accounted for (Eta2) ranged from 0.26 to 0.92 (averaging 0.76). For the NASCOM, DBT and CM sessions were significantly different on the subscales of Focus on feelings, Shared understanding, and CBT items (average Eta2 = 0.58), as well as on the total score. Across the two measures, the subscales that differentiated less (say, Eta2 ≤ .10) were either ones containing items universal to both therapies (from the DBT scale, Validation strategies and Irreverent strategies; and from the NASCOM, Facilitative conditions) or the subscales that had low frequency items (from the DBT scale, DBT 147 protocols, Exposure based principles, and Case management strategies; and from the NASCOM, Therapeutic ruptures, and Gaining insight). Table 3.4 Subscale and Total Score Profiles on the Two Adherence Scales DBT CM One-way Eta2 Adherence measure/Subscale sessions sessions ANOVA P (Va- (n=29) (n=27) F(1, 54) riance) DBT Reciprocal communication 4.29 3.99 18.85 <.001 0.26 strategies Validation strategies 4.21 4.34 4.74 .034 0.08 Contingency management 4.11 1.44 309.53 <.001 0.85 Problem assessment 4.00 1.95 304.49 <.001 0.85 strategies Problem solving / skills 3.93 1.42 655.29 <.001 0.92 training Irreverent strategies 3.91 2.99 5.96 .018 0.10 Dialectical strategies 3.91 1.53 350.43 <.001 0.87 Structural strategies 3.87 1.90 266.40 <.001 0.83 Cognitive based strategies 3.74 1.53 136.67 <.001 0.72 DBT protocols 0.48 0.00 3.89 .054 0.07 Exposure based principles 0.00 0.00 N/A Case management strategies 0.00 0.04 1.076 0.304 0.02 Total score 4.03 2.50 399.23 <.001 0.88 CM Therapeutic ruptures 0.07 0.15 1.26 0.267 0.02 CBT items 1.99 0.26 324.05 <.001 0.86 Gaining insight 0.72 1.01 3.26 0.077 0.06 Focus on feelings 0.63 1.45 52.26 <.001 0.49 Shared understanding 0.79 1.80 34.13 <.001 0.39 Facilitative conditions 3.62 3.62 0.00 0.984 0.00 Total score -1.41 5.11 141.17 <.001 0.72 3.2.3 NASCOM Adherence Cut-Off Figure 3.1 shows each session’s ratings on the NASCOM and the corresponding DBT adherence scale, demonstrating that the two treatments could be nearly perfectly differentiated. Apart from one DBT session outlier (relatively high on both adherence scales), and two CM sessions with NASCOM total scores of 1, there was clear differentiation of sessions based on the adherence scores. Consequently, within the current set of adherence ratings, a NASCOM total score of 2 or greater provided 92.6% 148 sensitivity and 96.6% specificity for correctly classifying the 27 CM modality sessions. In contrast, a score of 3.4 or greater on the DBT adherence scale provided 100% sensitivity and 100% specificity for correctly classifying the 29 DBT modality sessions. Figure 3.1. Individual Session Total Scores on NASCOM and DBT Adherence Scales 3.3 Discussion This chapter introduces a measure of adherence to CM called the NASCOM. Results presented provide preliminary evidence that the measure has acceptable psychometric properties. The NASCOM showed good inter-rater reliability across a number of key items used in CM, both within the CM treatment modality and when the CM and DBT treatment sessions were combined. These results are encouraging in 149 showing that psychodynamically oriented therapies, and specifically CM, can be successfully rated for adherence. However, a number of items had low inter-rater agreement. The lack of significant inter-rater agreement on several items in the NASCOM can possibly be explained by floor effects in the ratings for those items. It is a common finding among rating scales for the reliabilities to vary considerably at the item level (Barber, Liese, & Abrams, 2003). As outlined in the Results section, agreement was rarely outside one point of variation on the items that were rated as occurring in less than 5% of sessions. Freire, Elliot, and Westwell (2014) raise as a limitation of their scale and other therapy process measures that they often generalise “from relatively brief segments of therapy to therapist performance in general, while ignoring context and participant internal experiences” (p. 225). This is very relevant for psychodynamic therapies, as so much of the therapist response is dependent on context and accurate attunement to internal experiences, and may explain why for several items there was a floor effect. Alternately, it may be an issue with inadequate training, or the therapists may simply have not seen the benefit of such components. The two Facilitative conditions items, Warmth and Rapport, also had low inter- rater reliability and these have also struggled in other studies to reach acceptable inter-rater agreement (see Godfrey, Chalder, Ridsdale, Seed, & Ogden, 2007; Hill, O'Grady, & Elkin, 1992). These inter-rater reliability results potentially indicate that the instrument contains redundant items that could be removed in future and would make the scale easier to use. Discriminant analyses of the NASCOM items showed that CM could be clearly distinguished from DBT. These findings suggest that CM therapists used interventions 150 consistent with CM. Further, the data in Table 3.3 shows that proscribed items were rarely used by therapists undertaking CM. In developing the NASCOM, the authors attempted to follow Waltz et al.'s (1993) guidelines for adherence measures, including four types of items: (a) therapist behaviours that are unique to that treatment and essential; (b) behaviours that are essential to a treatment but not unique; (c) behaviours that are compatible with the specific modality, not prohibited but neither necessary nor unique; and finally (d) behaviours that are proscribed. We were able to include items covering (b)-(d), but we had difficulty with (a) as we could not identify any behaviours that were unique to CM. In fact, all therapist behaviours that are essential in CM are also components of DBT; however, the ratio of these behaviours would be expected to be higher in CM. Hence, it makes sense that it would be easier to discriminate CM from other therapies based on the absence of proscribed items rather than the presence of required items. Karterud et al. (2013) note in the development of their Adherence and Competence Scale for Mentalisation-based treatment that the difference between “essential” and “unique” was difficult to establish, since many psychotherapies attempt to promote exploration and curiosity, as well as a focus on affect. Their manual states that “the unique aspects of MBT lie less in each item than in the overall ‘package of items’, i.e., the total design” (p. 708). As is common in much of the research on development of adherence measures, a limitation in this study is the sample size. Another possible limitation related to sample size is the small number of sessions sampled for each client receiving therapy. As it was expected that a number of intervention components would be used only occasionally, more sessions may have been needed to capture these interventions. It is 151 possible that with a larger sample some additional intervention components may have achieved significant inter-rater reliabilities. Another limitation in the current study is that the raters knew some of the therapists being rated, and this may have influenced their ratings. Each of the raters evaluated the same therapist six times for therapists who delivered both models and three times for those who provided only one model. Hence, it is possible that the raters formed global impressions of therapists, which may have influenced their subsequent ratings. In the current study, only adherence was rated, not competence. Rating competence is costly, given that expert raters are required. Waltz et al. (1993) suggest that competence is a requirement in some situations and not others. They suggest that one of the circumstances in which competence can be presumed is “when expertise in all treatment conditions is equally represented by the investigative or supervisory team” (p. 628). This was the case for the current RCT, for which the NASCOM was developed. The rating of the extent to which a particular intervention component is delivered in an adherence scale, as was done with the NASCOM, opens up the possibility of empirically exploring aspects of the therapeutic alliance. Adherence ratings could be correlated with scores from instruments measuring the therapeutic alliance so as to explore how certain intervention components may be associated negatively or positively with the therapeutic alliance. The measurement of the extensiveness of intervention components can also be used to monitor what intervention components are employed at different stages of therapy. 152 This study identified a cut-off score of 2 or higher on the NASCOM in order for the session to be deemed “adherent” to CM. Further research should determine whether this cut-off score correlates with competence in the model. Beyond research studies, at a clinical level, the scale could also be used to provide feedback to therapists regarding aspects of the therapy that they use frequently and other aspects of the therapy that they are neglecting. 3.4 Conclusion The NASCOM was developed specifically for CM, a treatment approach adapted specifically for clients with a diagnosis of BPD by Meares (2004, 2005). It was evaluated on sessions from patients with a diagnosis of BPD. As such, it is unclear if the NASCOM is only valid for testing adherence for BPD or also for other disorders. However, there is no reason to expect that the NASCOM would not be able to be generalised to other disorders. As there were similarities between the NASCOM and the SPRS, especially in relation to the items reflecting Exploratory Therapy in the SPRS, it may have been useful to have rated the sessions in this study against the SPRS as well as the NASCOM. Such a comparison of the two instruments may have shed light on the question raised by Perepletchikova et al. (2007) as to whether new adherence scales need to be created and validated each time a new or revised therapy is tested for adherence. In summary, the NASCOM was developed to assess adherence to CM, a psychodynamically oriented treatment. It was developed in the absence of an existing scale with demonstrated good psychometric properties in order to discriminate CM from DBT in an RCT of treatments for BPD conducted in Newcastle, Australia. The NASCOM was found to have good inter-rater reliability. These results should be 153 considered preliminary—the findings are based on one un-replicated sample of therapists. A much larger sampling can now take place to test adherence to CM in outcomes studies of CM. The NASCOM was also able to distinguish CM from DBT in the current RCT. However, the items reflecting DBT practice also reflect cognitive behaviour therapies generally and on that basis could be used in other trials between CM and CBT therapies. While there are challenges in developing adherence scales for psychodynamic therapies, the development and testing of the NASCOM provides yet another small step in redressing the limited number of adherence studies for psychodynamic therapies. 154 Chapter 4 Dialectical Behaviour Therapy versus Conversational Model for Borderline Personality Disorder: Randomised Clinical Trial in a Public Sector Mental Health Service Chapter 1 outlined the need for an RCT comparing DBT and CM. Chapter 2 described the methodology of the RCT. Chapter 3 outlined an adherence tool developed to measure adherence to CM by therapists in the RCT. Chapter 4 reports on the results of this RCT in terms of the two co-primary outcomes of i) suicide attempts and NSSI and ii) depression severity, as well as a range of secondary outcomes. A more condensed version of the information described in Chapter 2 regarding the background to the study and the methodology is included here for the reader’s ease. 155 Borderline Personality Disorder (BPD) is associated with a high degree of suffering, high rates of suicide attempts (Lieb et al., 2004), and a lifetime suicide mortality rate of approximately 10% (Black et al., 2004). There are a number of psychotherapies that have been developed specifically for the treatment of BPD, including Dialectical Behaviour Therapy (DBT) and the Conversational Model (CM). DBT is a cognitive behavioural treatment developed specifically to treat BPD, targeting the development of skills to build a “life worth living” (Linehan & Wilks, 2015) and has been the focus of more clinical trials than any other psychotherapy for BPD (Cristea et al., 2017). Across these trials, results generally show a reduction in suicidal and non- suicidal self-injurious episodes, inappropriate anger, and service utilisation. There are two high-quality studies, adequately powered by large sample sizes and where DBT was delivered with fidelity, which have compared DBT with another active treatment. Linehan et al. (2006b) compared DBT with “treatment by experts” that included clinicians in the community with expertise in treating BPD using treatments other than DBT. DBT was found to be superior on outcomes of suicidal and non-suicidal self- injury, treatment retention and service utilisation. McMain et al. (2009) compared DBT with Generalised Psychiatric Management (GPM) and found that both treatments demonstrated significant reductions in suicidal and non-suicidal self-injury (NSSI) between pre-treatment and post-treatment, as well as on a range of other clinically relevant measures. CM is a psychodynamic model of treatment developed specifically to treat BPD, targeting the development of a healthy sense of self (Meares, 2005). It focuses heavily on the therapeutic relationship as a template for other relationships and aims to help individuals increase their capacity for reflective functioning. CM has been evaluated for 156 BPD in a published study (Stevenson & Meares, 1992), using a pre-post design, with a replication study (Korner et al., 2006), using a treatment as usual waiting list control, but has not yet been tested in a randomised trial. Both CM studies showed significant reductions in the number of diagnostic criteria met and reduction in suicide and non- suicidal self-injury, violent behaviour, and hospital admissions after one year of therapy, with gains maintained at five-year follow-up (Stevenson et al., 2005). There is no direct evidence from comparisons of active models developed specifically for the treatment of BPD that any one form of psychotherapy is superior to any other model. In the most recent Cochrane review of BPD (Stoffers et al., 2012), the authors outline several limitations to the studies of outpatient psychotherapeutic treatment for BPD. Many have small sample sizes (range n = 47 to 180). Apart from DBT, most treatment models have only been evaluated in one or two studies (Stoffers et al., 2012) and the majority of studies have been conducted by investigators who developed the treatment or who have a strong allegiance to one particular model (Bateman & Fonagy, 2009; Farrell et al., 2009; Gregory et al., 2008; Levy et al., 2006; Linehan & Wilks, 2015). Among studies in which all of the investigators have allegiance to one particular model, outcomes have been consistently in support of the treatment model to that model (Luborsky et al., 1999). Hence, replications are needed, particularly by independent researchers not involved in treatment development. Most studies have been compared against treatment as usual, and a positive finding for the active treatment might, therefore, be expected (Budge et al., 2013). Further, most have been conducted in university settings by highly trained therapists, and it is unclear how well the effects would generalise to real world clinical settings (Roy-Byrne et al., 2003). 157 The aims of the current study are to evaluate DBT in a routine clinical setting and to compare CM against another therapy for BPD with an established evidence base. By comparing DBT with CM, the present study attempts to address some of the limitations identified above. The study has a large sample size, utilises intention to treat analyses, and has incorporated measures of fidelity to treatment. The researchers are not the treatment developers and do not work alongside the treatment developers. Most importantly, the principal investigators have allegiance to both therapeutic models and the study was carried out in a real world setting with clinicians employed in public sector mental health services, who are usually responsible for implementing both these therapies. 4.1 Method 4.1.1 Study Design This was a single site, two-armed parallel randomised clinical trial (RCT) designed to investigate the effectiveness of CM and DBT in a public sector mental health service in Australia. The sample population comprised adults with a primary diagnosis of BPD and recent suicide attempts and/or NSSI episodes. The main aim of this study was to compare CM with DBT for two co-primary outcomes at post- treatment (14 months): change in the number of combined endpoint episodes of suicidal and non-suicidal self-injuries; and change in depression severity. We expected that: (1) both treatments would lead to significant change after 14 months; (2) DBT would be more effective in reducing the number of episodes of suicidal and NSSI after 14 months; and (3) CM would be more effective in reducing depression severity after 14 months. The study was approved by the Hunter New England Human Research 158 Ethics Committee (Reference Number: 06/12/13/5.11) and registered with the Australian New Zealand Clinical Trials Registry (ACTRN 12612001187831). 4.1.2 Participants This study was conducted at the Centre for Psychotherapy, a specialist outpatient service for BPD and/or Eating Disorders of the Hunter New England Mental Health Service (a public sector Mental Health Service), located in Newcastle, New South Wales, Australia. Referrals were accepted from community mental health teams, general practitioners or private therapists. As part of the Australian public health care system, there were no treatment costs for patients attending the Centre. Recruitment started in January 2007, and the final sample size of 162 was reached in April 2013. The final post-treatment data (14 months) collection occurred in June 2014. The inclusion criteria for this study were: (1) BPD according to DSM-IV criteria (American Psychiatric Association, 1994)—with the diagnosis of BPD made by a consultant psychiatrist using the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (SCID-II; First et al., 1995); (2) three episodes of suicide attempts and/or NSSI in the past 12 months assessed by a consultant psychiatrist during psychiatric interview, where non-suicidal self-injurious behaviour is defined in accordance with the Suicide Attempt and Self-injury count (Linehan & Comtois, 1996) as acute, intentional self-injurious behaviour such as cutting, overdosing, burning and deliberate self-poisoning without the intention of dying; and (3) aged between 18 and 65 years. These inclusion criteria were the same as the inclusion criteria for the service, and as such, the study sample was representative of the patients generally seen at the service. The exclusion criteria for this study were: disabling organic conditions; current acute psychotic illness; antisocial behaviour that 159 poses a significant threat to staff and fellow patients; developmental disability; living more than one hour’s drive from the treatment centre; inability to speak or read English; current substance dependence other than nicotine (as measured by SCID-I; but eligible for entry once no longer meeting the criteria for dependence); and prior treatment with DBT or CM (as reported by the patient). 4.1.3 Procedure Standard practice at the service was that upon receiving the referral, patients were allocated an appointment for an initial assessment with a psychologist, social worker, occupational therapist or mental health nurse therapist (all with a minimum of five years mental health professional experience) to determine whether they met eligibility criteria for the service and if there were any significant barriers that would interfere with their ability to make use of therapy. If at this assessment patients appeared to meet criteria for the service and were interested in pursuing treatment, they were assigned an appointment for a diagnostic interview with a consultant psychiatrist. Following diagnostic assessment, all patients meeting inclusion criteria were invited by a research assistant to participate in the study. Patients were provided with a full explanation of the procedures and treatment conditions, including the stipulation that because it is a randomised trial, they would not be permitted to change treatment models during the trial. After being provided with a complete description of the study, written informed consent was obtained from all participants. Participants received AUD $20 per assessment to cover transport and related costs. Participating in research assessments was not obligatory to receiving treatment. Those patients who did not wish to consent to the study received the same treatment 160 as those who consented to participate (either DBT or CM determined by patient preference and therapist availability). Consenting participants were randomly allocated to either DBT or CM. Figure 4.1 illustrates the flow of participants in the RCT. 161 269 met inclusion criteria for initial assessment Following referral to the service 103 excluded: 45 did not meet inclusion criteria (8 did not meet full diagnostic criteria for BPD, 3 did not meet self-harm criteria, 9 had interfering comorbid disorders, 21 previous DBT, 2 previous CM, 2 other reasons) 58 declined to participate (19 wanted DBT, 8 wanted CM, 4 not able to attend due to work commitments, 27 other reasons) 166 completed Baseline assessment T0 & randomised 83 allocated to Dialectical Behaviour Therapy 83 allocated to Conversational Model 2 did not start intervention 2 did not start intervention 81 started intervention 81 started intervention 36 discontinued intervention 33 discontinued intervention 28 (35%) lost to follow-up 20 (25%) lost to follow-up 10 refused to complete assessment 5 refused to complete assessment 18 uncontactable 14 uncontactable 1 was not approached for assessment (due to violent threats against staff) 81 analysed utilising “intention to treat” analysis 81 analysed utilising “intention to treat” (T0 – data complete, 22 missing T1 analysis data, 28 missing T2 data) (T0 – data complete, 14 missing T1 data, 20 missing T2 data) Figure 4.1. Consort Diagram DBT= Dialectical Behaviour Therapy; CM= Conversational Model; T0 = baseline, T1 = 7 months of treatment (mid-therapy), T2 = 14 months of therapy (post-therapy) 162 At the request of the Hunter New England Human Research Ethics Committee, a stratified randomisation procedure was used to maximise the likelihood of comparable distributions across groups for gender and current treatment status for prescribed antidepressants (prescribed or non-prescribed). A computerised formula with blocked randomisation (blocks of 4) was used by an independent research manager of the health service. A sealed, opaque envelope containing randomisation status was given to participants at the end of their assessment, so that participants, therapists, and researchers were blind to treatment allocation prior to participants opening the envelope. After randomisation, the waiting list manager in the service allocated participants to a therapist. Participants started therapy within two weeks of randomisation. As is consistent with psychotherapy trials for BPD, medication was not standardised; its type and amount were decided on an individual basis by participants’ medication providers (general practitioners or psychiatrists). 4.1.4 Assessments Data was collected at six time points: at baseline (T0); at the mid-treatment mark of seven months (T1); at post-therapy at 14 months (T2) and at 26 month (T3), 38 month (T4) and 74 month (T5) follow-up. Only time points T0, T1 and T2 will be reported in this thesis. Follow-up data (T3-T5) are still being collected and while it will not be reported here, it is noted that this is occurring given the importance of collecting longer-term follow-up data for outcome research in BPD. Experienced research assistants with a minimum bachelor’s degree in psychology conducted all research assessments, blind to treatment allocation status. Although the research assessments were conducted in the same building where 163 therapists treat patients, the research assistant was scheduled to work at different times to when the skills training groups were run in order to maintain blindness. Moreover, the assessors regularly reminded both participants and therapists that they needed to be kept blind to treatment allocation. 4.1.4.1 Primary outcome measures. There were two co-primary outcomes: change from baseline to post-treatment in (1) number of episodes of suicidal and NSSI and (2) self-reported depression severity scores. While the primary outcome comparison was baseline to post-treatment, two other planned comparisons were also assessed: change from baseline to mid- treatment and from mid-treatment to post-treatment. The specific measures used to evaluate each outcome in the study are outlined below with the outcome domain identified first, followed by the measure used. Further detail of each of the measures is included in Chapter 2. (a) Combined outcome of any episode of suicidal and non-suicidal self-injury: The Suicide Attempt and Self-Injury Count (SASI-Count; Linehan & Comtois, 1996; Linehan, Comtois, & Lungu, 2011). This is a briefer version of the Suicide Attempt Self Injury Interview (SASII; Linehan et al., 2006a), which assesses self-injury in more detail. (b) Depression severity: Beck Depression Inventory II (BDI-II; Beck et al., 1996). 4.1.4.2 Secondary outcome measures. (a) BPD Severity: Borderline Personality Disorder Severity Index (BPDSI-IV; Arntz et al., 2003) (b) Interpersonal Problems: Inventory of Interpersonal Problems (IIP; Horowitz, Rosenberg, Baer, Ureno, & Villasenor, 1988) 164 (c) Dissociation: Dissociative Experiences Scale (DES; Bernstein & Putman, 1986) (d) Sense of self: Sense of Self Inventory (SSI; Basten, 2008) (e) Mindfulness: Kentucky Inventory of Mindfulness Skills (KIMS; Baer, Smith, & Allen, 2004) (f) Emotion Regulation: The Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004) 4.1.5 Treatment Patients were required to commit to 14 months of twice weekly psychotherapy in either program: twice weekly individual therapy sessions in the CM condition, and a once weekly individual therapy session and once weekly skill training group session in the DBT condition. 4.1.5.1 Dialectical Behaviour Therapy (DBT). DBT is a manualised treatment (Linehan, 1993a; Linehan, 1993c) that combines treatment strategies from behavioural, cognitive, and supportive psychotherapies. It includes weekly individual pre-treatment sessions for approximately four weeks (variable depending on when there is an available entry into the skills training group). Pre-treatment sessions include orientation to the treatment model, exploring goals and eliciting a commitment to therapy. Pre-treatment sessions are followed by concurrent weekly individual and skills training group sessions for 12 months. The individual therapy sessions take approximately one hour per week and apply directive, problem-oriented techniques (including behavioural skill training, contingency management, and cognitive modification) alongside supportive techniques. The skills training group meet weekly for 2.5 hours and follow a psycho-educational format. 165 Behavioural skills in three main areas are taught in three modules as follows: interpersonal effectiveness, distress tolerance, and emotion regulation skills. A two- week focus on mindfulness, which is the core skill taught over the entire course, precedes each module. The skills group includes the teaching and application of skills, and their practice between sessions. In addition to attending individual therapy and skills training groups, participants had access to telephone coaching with their therapist during working hours. Outside of therapist working hours, in our service, participants had access to a telephone service staffed by a roster of six DBT therapists from the service, which was available seven days per week from 8:30 a.m. to 10:00 p.m. DBT therapists attended a weekly consultation team meeting designed to provide support and assist therapists to be adherent to the treatment model. The 14-month schedule allowed for the pre-treatment phase in DBT to occur within the 14-month period before participants commenced 48 sessions of skills training that occurred concurrently with the individual therapy. The 48 sessions usually took an entire year taking into account that on occasional weeks the skills groups did not run (due to public holidays, the service being closed for two weeks over Christmas or therapists being on leave). 4.1.5.2 Conversational Model (CM). Participants in the CM treatment arm attended twice weekly individual therapy for 14 months. The treatment model was developed by Meares (Meares, 2004; Meares, 2005) and is outlined in a published treatment manual (Meares, 2012). The individual therapy sessions are approximately of one hour’s duration and are nondirective. The focus is on understanding the patient’s emotional experience 166 and actively describing that experience back to the patient. The therapist actively looks for subtle signs of emotionally misunderstanding the patient, leading to mutual self- reflection and repair of the moment of disconnection in the therapeutic relationship. High value is placed on the patient’s real experience (as against socially acceptable experience) and the development of an authentic personal narrative. The patient is encouraged to find links between the maladaptive relationship patterns they have developed in their current social world and the relationship pattern they have with the therapist (and possibly, but not necessarily, the links with their childhood relationships). Fourteen months of therapy was selected as the treatment length to match that being received in the DBT condition. Previous research on CM has involved treatment of one or two years’ duration. There was no explicit pre-treatment phase in CM; however, it did involve an extended assessment that occurred at the beginning of the 14 months. 4.1.6 Therapists Treatment was delivered by 32 therapists, all with a minimum of two years of clinical experience. Therapists providing the treatment were clinicians who were employed to work in the government health service and not hired specifically for this trial. Half of the therapists in the trial were employees of the Centre for Psychotherapy; some therapists were employed in community or hospital mental health settings and served as visiting therapists to the Centre for Psychotherapy. Disciplines included psychologists, psychiatrists or psychiatric trainees, social workers, mental health nurses and occupational therapists. Some therapists provided treatment in both models (n=12), some provide treatment in DBT only (n=7), and some in CM 167 only (n=13). Whether therapists elected to provide treatment in one or both models was the choice of the therapist, and this was not considered during randomisation. 4.1.6.1 Supervision. Therapists providing treatment in the DBT arm of the trial were required to complete a minimum of four days of DBT training. All had completed additional two- day training sessions with the treatment developer’s training company, Behavioral Tech (http://behavioraltech.org/index.cfm), and 14 of the therapists who delivered DBT completed 10-day intensive training in DBT in 2010 with Behavioral Tech. Therapists providing treatment in the CM arm of the trial attended an introductory program; for the allied health and nurse therapists, this involved a series of a minimum of 6 x two-hour seminars. For psychiatric trainees and psychiatrists, due to scheduling difficulties, the initial training was more individually tailored in meeting one on one with a staff specialist psychiatrist (2nd author) with extensive experience in the model to cover the seminar material. The majority of therapists in the CM arm of the trial had either completed or were completing a three-year part-time diploma specifically in CM with the treatment developer’s training organisation, Australian and New Zealand Association of Psychotherapy (ANZAP; http://www.anzapweb.com/). 4.1.6.2 Treatment fidelity. Fidelity was monitored through supervision. Therapists delivering DBT attended a weekly consultation team meeting, with approximately six members in each consultation team. Therapists who had only recently learned DBT also received individual supervision initially. For therapists providing CM, all therapists were involved in weekly supervision with an experienced CM supervisor either individually or in pairs. 168 Modality-specific adherence scales were used to evaluate treatment fidelity. All clinical sessions were audiotaped with the patient’s permission for the purpose of supervision and to formally assess treatment adherence. Adherence raters were blind to treatment allocation. A randomly selected 5% sample of all sessions were rated for adherence. DBT sessions were rated by a coder trained in the use of the Dialectical Behavior Therapy Global Rating Scale (Linehan, 1993b). An overall score of 4 or higher indicates that the session was conducted adherently. Reliability checks were conducted between the coder and a member of Linehan’s team. In the absence of a suitable adherence scale for CM, one was developed for use for this study. CM sessions were rated by a coder trained in the use of the Newcastle Adherence Scale for CM (NASCOM; Walton, Goldman, Bendit, & Startup, under review). The NASCOM is a 25-item scale with well-established psychometric properties that assesses the percentage of CM techniques used within the session. Sessions were coded by a rater, trained in its use by the developer of the scale, and reliability checks were conducted with the scale developer. Adherence coding commenced after the trial was initiated. It was decided that therapists would not receive the results of the adherence coding during the trial, as it is unrealistic that such feedback would generally occur in a real world setting and feedback provided mid-way through the trial would have introduced a confounding variable in terms of outcome. Therapists received adherence feedback once all participants recruited in the study had finished therapy. 169 4.1.7 Statistical Analysis 4.1.7.1 Sample size. The planned number of participants in the study was 162. This was based on the assumption that two-thirds would be retained at post-treatment (i.e., 54 participants per intervention condition). Family-wise Bonferroni-corrected statistical tests were employed throughout—namely alpha divided by 2 for the two co-primary outcome measures (i.e., 0.025), and alpha divided by 6 for the secondary outcome measures (i.e., 0.0083). During initial planning for the study, it was anticipated that the major analyses would comprise a series of linear multiple regressions with a small number of predictor variables (e.g., planned comparisons between conditions and some covariates). Consequently, it was expected that a sample of 162 participants would provide sufficient statistical power (80%) to detect modest population associations (e.g., simple or partial correlations) of 0.30 or higher for the primary outcome measures, using two-tailed statistical tests in the sample (or 0.34 or higher in the case of the secondary outcome measures). 4.1.7.2 Analysis. Analyses included a range of descriptive and analytical approaches, whereby all patients who attended a minimum of one treatment session after the baseline assessment and were randomised were included in the analysis. Consistent with an intention to treat approach, all of the available data from the three time points were included in each analysis (not just complete pairs), using specific planned comparisons to examine differences between time points (and account for the repeated measures component of the variance). For consistency, the major outcome measures were expressed as change scores from baseline (T0) i.e., subsequent phase of either mid- 170 therapy (T1) or post therapy (T2) minus baseline phase. Generalised Linear Modelling techniques (Generalised Estimating Equations, GEE) were used to examine differences between groups in the changes over time (with within-subject variation coded using subject IDs). For both primary and secondary outcomes, z scores were also calculated to show standardised change, expressed in relation to the grand SD of change. SPSS 22.0 software package for Window was used for statistical analyses. 4.1.7.3 Treatment of missing data. As shown in Figure 4.1, 28 participants were lost to follow-up evaluation in the DBT condition, and 20 participants were lost to follow-up evaluation in the CM condition. To take into account missing data and maximise usage of all available data, as noted above, the major analyses comprised a series of Generalized Linear Models (and, where appropriate, GEE analyses), following a similar approach to that adopted in a recent clinical trial reporting outcomes across multiple follow-up phases up to 36 months (Baker et al., 2014). Separate analyses are reported for each of the outcome measures, although a core set of predictors will be included in each analysis. 4.2 Results 4.2.1 Recruitment Participant flow across the trial is presented in Figure 4.1. In total, 269 patients were referred to the service during the recruitment period of the trial, of whom 45 did not meet inclusion criteria for the study and 58 declined to participate. A total of 166 people consented to randomisation and completed research assessments. Four of these did not attend the first therapy appointment, leaving 162 participants (see 171 Figure 4.1) with BPD who gave written informed consent, attended at least one therapy session, and were included in the study. 172 Table 4.1 Baseline Sociodemographic and Clinical Characteristics of Participants Statistical comparison a Overall DBT (n=81) CM or (n=162) (n=81) t-𝟐𝟐test p Sociodemographic variables 𝒙𝒙 Age, years: mean (s.d.) 26.6 (7.8) 25.8 (7.4) 27.3 (8.1) t =-1.260 .210 Female, n (%)b 125 (77%) 62 (77%) 63 (78%) Highest level of education, n (%) x2 = 4.444 .349 No high school certificate 49 (31%) 30 (37%) 19 (25%) Completed high school 27 (17%) 12 (15%) 15 (20%) Post-secondary, e.g., trade 33 (21%) 13 (16%) 20 (26%) Some University 31 (20%) 16 (20%) 15 (20%) University degree 18 (11%) 10 (12%) 8 (10%) Employed, n (%) 59 (37%) 30 (38%) 29 (37%) x2 = 0.002 .967 Ethnic origin, n (%) x2 = 0.084 .959 White Caucasian 139 (86%) 69 (85%) 70 (86%) Aboriginal 10 (6%) 5 (6%) 5 (6%) Other 13 (8%) 7 (9%) 6 (7%) English as first language, n (%) 157 (97%) 79 (98%) 78 (96%) x2 = 0.206 .650 Marital status, n (%) x2 = 3.575 .311 Single 95 (59%) 44 (54%) 51 (63%) Married/De-facto 54 (33%) 32 (40%) 22 (27%) Divorced 13 (8%) 5 (6%) 8 (10%) Clinical variables Lifetime Axis I Psychiatric Diagnoses Any anxiety disorder, n (%) 111 (69%) 54 (67%) 57 (70%) x2 = 0.258 .612 Any substance use disorder, n (%) 95 (59%) 50 (62%) 45 (56%) x2 = 0.636 .425 Current Axis I Psychiatric Diagnoses Major Depressive Disorder, n (%) 84 (52%) 43 (53%) 41 (51%) x2 = 0.099 .753 Any anxiety disorder, n (%) 89 (55%) 42 (52%) 47 (58%) x2 = 0.623 .430 Any substance use disorder, n (%) 45 (28%) 24 (30%) 21 (26%) x2 = 0.277 .599 Axis II Diagnoses (excluding BPD), 1.5 (1.4) 1.5 (1.5) 1.5 (1.3) t =0.433 .658 mean (s.d.) Psychotropic medication at baseline, 117 (73%) 56 (70%) 61 (76%) x2 = 0.795 .373 n (%) Antidepressant use at baselineb 99 (62%) 49 (61%) 50 (63%) a. Pearson’s chi-square or t-test was used depending on whether the variable was categorical or continuous. b. Both gender and antidepressant use at baseline were stratified. 4.2.2 Patient Characteristics Selected demographic and clinical characteristics of the two groups are presented in Table 4.1. Participants were mainly single, white Caucasian women in their 20s, with varying levels of education. The two groups showed no imbalances at baseline after randomisation with regard to sociodemographic and clinical characteristics. 173 4.2.3 Treatment Outcomes Intent to treat analyses showed a significant reduction over time in both co- primary outcomes of (i) any suicidal and/or NSSI (Wald’s X2 = 27.13, p<.001) and (ii) depression severity (Wald’s X2 = 142.02, p<.001). These results are detailed in Table 4.2. Given that there were differences in severity between suicide attempts and non- suicidal self-injury, they were also examined separately and are reported for the reader’s interest (see also Figures 4.2–4.4). They showed the same pattern of results. There were no deaths in either group. Compared to CM, DBT was associated with a significant differential benefit in depression severity scores from baseline to mid- therapy (Wald’s X2 = 8.05, P<.01) and from baseline to post-therapy (Wald’s X2 = 8.00, P<.01); see the group x time interaction effects in Table 4.2 and the associated mean changes displayed in Figure 4.5. 174 Table 4.2 Generalised Linear Model Results of Co-Primary Outcomes of Suicide Attempts and NSSI and Depression Severity Treatment DBT (n=81) CM (n=81) Time Effects b Group x Time Effects c Variable Mean (SD) Change Mean (SD) Change Wald’s X2 P Wald’s X2 P Z score Z score Co-primary outcome: Suicide attempts and NSSI (n=162) T0 Baseline 66.77 (109.30) 79.53 (114.38) T0 v T1: 20.99 <.001 G x T0 v T1: 0.34 .557 T1 7 mths 31.72 (69.68) [0.317] 39.72 (75.68) [0.410] T1 v T2: 8.03 .005 G x T1 v T2: 0.08 .935 T2 14 mths 10.10 (35.95) [0.496] 15.00 (30.09) [0.580] T0 v T2: 27.13 <.001 G x T0 v T2: 0.16 .688 Suicide attempts (n=79) a T0 Baseline 23.68 (55.10) 12.21 (29.39) T0 v T1: 6.42 .011 G x T0 v T1: 0.80 .371 175 T1 7 mths 3.14 (9.33) [0.270] 2.21 (4.21) [0.134] T1 v T2: 0.11 .743 G x T1 v T2: 0.14 .713 T2 14 mths 1.04 (2.89) [0.309] 2.72 (15.20) [0.133] T0 v T2: 5.41 .020 G x T0 v T2: 0.79 .372 Non-Suicidal Self-Injury (n=142)a T0 Baseline 64.11 (92.31) 81.94 (115.08) T0 v T1: 21.06 <.001 G x T0 v T1: 0.99 .319 T1 7 mths 33.35 (69.42) [0.270] 40.85 (78.84) [0.412] T1 v T2: 8.16 .004 G x T1 v T2: 0.00 .998 T2 14 mths 10.15 (37.49) [0.464] 14.91 (28.43) [0.606] T0 v T2: 28.89 <.001 G x T0 v T2: 0.55 .457 Co-primary outcome: Depression severity (n=162) T0 Baseline 38.63 (10.31) 35.64 (10.68) T0 v T1: 101.92 <.001 G x T0 v T1: 8.05 .005 T1 7 mths 23.08 (12.36) [1.019] 26.61 (13.73) [0.582] T1 v T2: 24.27 <.001 G x T1 v T2: 0.66 .416 T2 14 mths 15.94 (14.52) [1.479] 22.13 (17.78) [0.912] T0 v T2: 142.02 <.001 G x T0 v T2: 8.00 .005 ; T0 = Baseline T1 7 mths (mid-treatment); T2 14 mths (post-treatment). Significance levels for primary outcomes are 0.025 (Bonferroni-correction for 2 co-primary outcomes) a Suicide attempts and Non-Suicidal Self-Injury (NSSI) are included as sub-analyses of the aggregated co-primary outcome of Suicide attempts and NSSI. For the sub-analysis of suicide attempts and NSSI, only those with those behaviours at baseline were included in the analyses. N is included above. b Effects for time only: T0 v T1: Baseline-Mid treatment; T1 v T2: Mid-post-treatment; T0 v T2: Baseline-post-treatment c Effects for interaction by group and time: G x T0 v T1: Group x time effect for Baseline-Mid treatment; T1 v T2: Group x time effect for Mid-post- treatment; T0 v T2: Group x time effect for Baseline-post-treatment 100 80 DBT 60 CM 40 & NSSI 20 0 Baseline Mid Post Number of Suicide Attempts Assessment Timepoint Figure 4.2. Mean No. of Suicide Attempts & NSSI Across T0-T2 Time Points by Condition 25 20 DBT 15 CM 10 5 0 Baseline Mid Post Number of Suicide Attempts Assessment Timepoint Figure 4.3. Mean No. of Suicide Attempts Across T0-T2 Time Points by Condition 100 - 80 DBT 60 CM Suicidal Self 40 Injuries 20 0 Baseline Mid Post - Non Number Assessment Timepoint Figure 4.4. Mean Number of Episodes of NSSI Across T0-T2 Time Points by Condition 176 40 DBT 30 CM 20 10 Depression Scores 0 Baseline Mid Post Assessment Timepoint Figure 4.5. BDI-II Severity Scores Across T0-T2 Time Points by Condition Consistent with the findings on the co-primary outcomes, analyses of all secondary outcomes showed significant improvement over time (see Table 4.3). In our examination of differences in the group x time interactions, there were no significant differences between DBT and CM on changes in BPD severity, interpersonal problems, dissociation, or sense of self. DBT showed a significant differential benefit between baseline to mid-therapy (Wald’s X2 = 11.761, P<.01) and baseline to post-therapy (Wald’s X2 = 8.162, P<.01) in improving mindfulness skills and from baseline to post- therapy in changes in emotion regulation skills (Wald’s X2 = 7.035, P<.01). 177 Table 4.3 Generalised Linear Model Results of Secondary Outcomes Treatment DBT (n=81) CM (n=81) Time Effects a Group x Time Effects b Variable Mean (SD) Z score Mean (SD) Z Score Wald’s X2 P Wald’s X2 P BPD Severity T0 Baseline 40.13 (11.45) 36.80 (12.50) T2 14 mths 19.0 (9.83) [1.333] 19.49 (11.10) [1.232] T0 v T2: 198.55 <.001 G x T0 v T2: 0.582 .445 Interpersonal Problems T0 Baseline 128.43 (29.25) 120.38 (31.81) T0 v T1: 14.866 <.001 G x T0 v T1: 6.487 .011 T1 7 mths 111.90 (36.12) [0.455] 116.62 (30.99) [0.093] T1 v T2: 26.523 <.001 G x T1 v T2: 0.052 .819 T2 14 mths 94.25 (44.57) [0.903] 99.08 (39.89) [0.582] T0 v T2: 49.212 <.001 G x T0 v T2: 2.298 .130 178 Dissociation T0 Baseline 30.50 (18.14) 32.31 (18.02) T0 v T1: 7.278 .007 G x T0 v T1: 0.064 .800 T1 7 mths 24.85 (15.21) [0.257] 28.55 (18.99) [0.213] T1 v T2: 35.000 <.001 G x T1 v T2: 0.044 .834 T2 14 mths 17.31 (12.52) [0.684] 22.35 (19.31) [0.671] T0 v T2: 50.876 <.001 G x T0 v T2: 0.005 .946 Sense of Self T0 Baseline 72.56 (10.77) 70.41 (10.45) T0 v T1: 33.828 <.001 G x T0 v T1: 1.090 .296 T1 7 mths 63.50 (11.82) [0.604] 64.08 (14.28) [0.420] T1 v T2: 49.962 <.001 G x T1 v T2: 0.474 .491 T2 14 mths 52.38 (12.86) [1.134] 55.33 (18.18) [1.004] T0 v T2: 100.524 <.001 G x T0 v T2: 1.796 .180 Mindfulness T0 Baseline 97.46 (16.77) 99.38 (17.93) T0 v T1: 30.456 <.001 G x T0 v T1: 11.761 .001 T1 7 mths 114.09 (23.43) [0.688] 104.08 (20.35) [0.178] T1 v T2: 16.635 <.001 G x T1 v T2: 0.071 .791 T2 14 mths 123.88 (24.81) [1.083] 111.96 (23.93) [0.525] T0 v T2: 61.706 <.001 G x T0 v T2: 8.162 .004 Emotion Regulation T0 Baseline 134.34 (21.57) 133.53 (21.10) T0 v T1: 50.263 <.001 G x T0 v T1: 3.460 .063 T1 7 mths 110.55 (24.94) [0.795] 118.32 (25.21) [0.464] T1 v T2: 39.171 <.001 G x T1 v T2: 2.219 .136 T2 14 mths 87.08 (28.0) [1.502] 105.16 (33.90) [0.900] T0 v T2: 111.818 <.001 G x T0 v T2: 7.035 .008 ; T0 = Baseline T1 7 mths (mid-treatment); T2 14 mths (post-treatment). Significance levels for secondary outcomes are 0.0083 a (Bonferroni-correction for six secondary outcomes). Effects for time only: T0 v T1: Baseline-Mid treatment; T1 v T2: Mid-post- b treatment; T0 v T2: Baseline-post-treatment. Effects for interaction by group and time: G x T0 v T1: Group x time effect for Baseline- Mid treatment; T1 v T2: Group x time effect for Mid-post-treatment; T0 v T2: Group x time effect for Baseline-post-treatment 4.2.4 Comparison of Treatment Completers and Those Who Dropped Out Of those assigned to DBT, 45 (56%) completed the full 14 months of treatment. Of those assigned to CM, 48 (59%) completed the full 14 months of treatment. Those who did not continue with treatment were encouraged to attend for assessments. We compared the results of those who dropped out of treatment with those who completed treatment, regarding baseline variables, in terms of sociodemographic and clinical variables and baseline scores of outcome variables and found no significant differences with 2 exceptions. Those who dropped out, were significantly lower in age (Mean of 25.16, SD 6.51) than those who completed treatment (Mean of 27.64, SD 8.43; (F1, 160) =3.712, p<.05) and had significantly lower scores on the Sense of Self Inventory (Mean of 68.65, SD 10.73; indicative of a stronger sense of self) than those who completed treatment (Mean of 73.02, SD 10.35; (F1, 107) =3.712, p<.05). 4.2.5 Fidelity Adherence scores were calculated for 218 DBT sessions (approximately 5% of all clinical sessions). The average adherence score on the DBT Adherence Rating Scale was 4.1 (Linehan, 1993b) and 77% of sessions had a score of 4.0 or above (the cut-off for adherence on the DBT Adherence Rating Scale). 174 sessions of CM were coded. The average was 3.9 and 99% of sessions had a score of 2 or above (the cut-off for adherence on the Newcastle Adherence Scale for Conversational Model). For CM, sessions could be clearly discriminated from DBT sessions. As reported in Chapter 3, for the 12 therapists providing both DBT and CM, 30 sessions of DBT and 30 sessions of CM were assessed on the scale for the alternative therapy, e.g., DBT sessions were rated on the CM adherence scale and vice versa. All 60 sessions were identified as non- 179 adherent, which confirms that therapists were delivering the treatment that the patient was randomised to and not blending the therapy with the other treatment. CM sessions scored on the DBT Adherence Rating Scale obtained an average score of 2.5 (SD = 0.37). DBT sessions scored on the NASCOM obtained an average score of -1.41 4.3 Discussion The current study is the first “real-world” comparison of DBT and CM for reduction of suicidal and NSSI episodes and depression severity among persons with BPD. It aims to extend our knowledge about how well DBT translates to public mental health treatment settings, with therapists in this trial being those that are routinely employed by the health service and with minimal restrictions to participant inclusion. Although CM has been the focus of a number of treatment studies, none have involved a randomised design or an active comparative treatment. Thus, the appropriate next step for investigation was to compare CM against an established active treatment using a randomised design. As expected, there was a significant overall reduction on the co-primary outcome variable over time for DBT and CM. The other two hypotheses were not supported. It was predicted that DBT would reduce suicidal and NSSI more than CM. However, the results demonstrated that both DBT and CM were associated with significant reductions in suicidal and non-suicidal self-injury, with no significant differences between the two therapies. This is surprising, as DBT deliberately and explicitly targets suicidal and non-suicidal self-injury, whereas CM does not. These results are consistent with the findings from the CM studies that suicidal and NSSI 180 improves without direct targeting (Korner et al., 2006; Stevenson & Meares, 1992; Stevenson et al., 2005). Whether these models work through common factors that have not been identified by the model developers or have differential change mechanisms that turn out to be equivalent cannot be determined from this study. However, this result is also in keeping with a previous RCT comparing DBT and a BPD adapted psychodynamic treatment (McMain, Guimond, Streiner, Cardish, & Links, 2012; McMain et al., 2009) that found no significant differences between the two active treatments. It was also hypothesised that CM would be associated with greater improvements in depression severity scores than DBT. Although both interventions were associated with significant reductions in depression severity scores over time, DBT was associated with statistically significantly lower scores than CM across the treatment year. This was also surprising, as previous studies of DBT have shown inconsistent impacts on depression severity scores (Stoffers et al., 2012). In contrast, CM in a previous wait list control study (Stevenson & Meares, 1992) showed substantial changes in depression severity, as did Mentalisation Based Therapy, another psychodynamic treatment for BPD (Bateman & Fonagy, 2009). At face value, it is interesting to consider why DBT’s performance was superior to CM in reducing depression severity. One possibility is that DBT may have inherently more behavioural activation and behavioural activation therapy has a large amount of evidence in support of reducing depression (Ekers et al., 2014). This result provides evidence for DBT being able to impact on the borderline patient’s internal experience beyond purely behavioural problems. 181 Similarly to the results for suicidal and non-suicidal self-injury, for a range of secondary outcomes, in the domains of BPD severity, dissociation, interpersonal difficulties and sense of self, both treatment conditions contributed to significant improvements over time, with no significant differential changes between the therapies. Each of these changes is clinically important. Dissociation is regarded as a major challenge in treatment (Kleindienst et al., 2016) and for healthy functioning in the borderline patient’s social world. Interpersonal functioning improvements enable patients to assert themselves and to develop healthy friendships and intimate relationships. Symptom severity is a broader measure of dysfunction and distress than our primary outcomes of suicidal and NSSI and depression. And finally, sense of self is regarded as one of the critical components lacking for patients with BPD and has been reported by patients with BPD as one of the most debilitating and alienating symptoms they experience (Korner, Gerull, Meares, & Stevenson, 2008). DBT was superior to CM in improving mindfulness and emotion regulation scores. This is less surprising since these constructs are explicitly taught in DBT with psychoeducation, specific skills and continued practice (Linehan, 1993c) and the wording in these measures uses the same language as that taught in DBT (Baer et al., 2004; Gratz & Roemer, 2004). CM does not develop the language of emotion regulation and mindfulness, and therefore the instruments using these linguistic constructs are likely to favour DBT. However, it does suggest that in DBT, two key areas explicitly targeted as skills deficits do change in response to treatment. The dropout rates in this study were comparable across the two treatments (41% for CM and 44% for DBT). Treatment dropout rates from individuals with BPD are higher than from those with Axis I disorders (Kelly et al., 1992). Regardless, dropout 182 rates in this study were higher than in the efficacy studies (16% in Linehan’s (1998) original study, and 25% in Linehan et al. (2006b)), but similar to other independent trials of DBT (43% in Clarkin et al. (2007), 37% in Verheul et al. (2003) and 38% in McMain et al. (2009)). In Australia, alternative publicly funded treatments for BPD are relatively available and accessible. Study participants received psychotherapy at no cost and were free to choose another therapist if they were not satisfied with the one to whom they were assigned. This is similar to the NHS in the UK, where other effectiveness studies of DBT have also shown higher dropout rates (Feigenbaum et al., 2012; Priebe et al., 2012). 4.3.1 Strengths and Limitations There are a number of strengths to the current study. The service had already been the setting of an RCT comparing six months of DBT with waiting list control (Carter et al., 2010). As such, the feasibility of running an RCT at this service had already been established. The sample size for the current trial was sufficiently large, relative to most previously published RCTs in the field of BPD, and provides adequate power to examine differential changes across the primary and secondary outcomes. Additionally, outcomes were assessed using well-validated measures commonly used in other BPD trials. Stoffers et al. (2012) in their recent Cochrane review of psychological therapies for BPD suggested “there is an urgent need for independent research endeavours” (p. 77), and the current trial goes some way to providing such evidence, given that none of the authors is a treatment developer, nor have they worked clinically alongside the treatment developers. The co-primary investigators (including the author of this thesis) are both trained in and deliver both models of therapy, and their allegiance to treatment model (evidenced in their training history) 183 provides a counter-balance to prevent overall researcher allegiance (Luborsky et al., 1999). Several elements in the trial were designed to reduce potential bias. The stratified randomisation based on gender and antidepressant medication allowed us to reduce outcome biases that could occur if uneven allocation occurred in these two areas following (unstratified) randomisation. Blindness of raters during the assessment of outcome variables and for adherence ratings further reduces potential biases. By measuring adherence in both models, cross-contamination of one model by the other could be excluded. Stoffers et al. (2012) have criticised the majority of trials of psychological therapies for BPD on the basis of the amount of professional contact, in that control group participants did not receive comparable amounts of professional attention. In the current study, participants in the CM condition attend twice weekly individual therapy, while those in the DBT condition attended once weekly individual therapy and once weekly skills training. As such, the amount of attention received from professionals was comparable, even though those in the CM group received two hours of face-to-face individual therapy and those in DBT received one hour of face-to-face individual therapy and 2.5 hours of skills training (with two therapists and usually eight people in a group). Some weaknesses arise from the study being completed in a real world clinical setting. Ideally, a third arm, involving a minimal treatment control group, would have been included. Without a control group, we are unable to eliminate the possibility that changes that occur are a result of exposure to common external factors or regression to the mean. However, it was deemed unethical for persons at known risk of suicide to 184 be denied an evidence-based psychotherapeutic treatment when one was available. Further, in the case of DBT, there was already evidence of efficacy, given it has been compared to a control group a number of times, as shown in Table 1.2. As such, the significant results found in the current study are greater than what would have been expected if there had been no intervention; for example, in the DBT condition there was a standardised (z-score) change in depression severity from baseline to post- treatment of 1.48, and 0.91 in the CM condition (see Table 4.2), which clearly exceed the benefits previously reported for minimal treatment control conditions as shown in Table 1.3. Many of the measures used in the study were self-report and may result in reporter bias. It would be preferable to have more clinician-administered measures; however, we used the self-report measures because of limited resources. Whilst the Research Assistants were blind to treatment condition and they verbally reported that they had remained blind, there was no formal check in relation to this. While we have worked hard to reduce potential areas of bias, some remain. In most clinical trials, therapists only deliver one of the interventions under evaluation or all therapists deliver both. In the current study, some therapists deliver only one of the models, and some therapists delivered both. This is in keeping with what would frequently occur in routine clinical settings but introduces some potential therapist bias. Although accessing any other psychotherapy outside of the trial was prohibited, there was no control of psychiatric medications. The two treatment arms were stratified for antidepressant use at baseline, as one of the primary outcomes was changes in depression severity scores. Other psychiatric medications were not 185 controlled. However, as several guidelines for treatment of BPD (National Health and Medical Research Council, 2012; National Collaborating Centre for Mental Health, 2009) report that there is little evidence that medication treatment in BPD is effective, this is not likely to be an important confound, as it would be in treatment trials evaluating Axis I psychiatric disorders. In this study, a number of participants continued psychotherapy after the one- year study period, as has been the case in previous RCTs with BPD (Bateman & Fonagy, 2009; Doering et al., 2010; Giesen-Bloo & Arntz, 2007). Although the Centre where the study is being undertaken sits within public sector mental health services, it is a specialist service for BPD and Eating Disorders. As such, while the therapists were not specifically hired for the study and were more representative of therapists found in real world settings, they do have expertise in BPD beyond that which is likely to be found in a general community service; this may limit the generalisability of the findings to general treatment settings. 4.3.2 Future Research Implications This is one of the few effectiveness trials in BPD and adds to the evidence pool, demonstrating that therapies that have been shown to be efficacious in tightly controlled settings can be applied in standard clinical settings with good outcomes. Further, this study adds to the growing evidence base that a variety of psychotherapy models, when adapted to working with patients with BPD, are clinically effective, resulting in improvement in a range of domains. Subsequent papers will report on one- year, two-year, and five-year follow-up data. These will be naturalistic, as there was no prohibition on subsequent treatments for the participants of this trial. 186 The next step in the research is to investigate whether we can improve outcomes by tailoring specific models to specific patients, answering the critical question: what works best for whom? 187 Chapter 5 Comparing the Therapeutic Alliance across Dialectical Behaviour Therapy and the Conversational Model in the Treatment of Borderline Personality Disorder Chapter 4 reported on the primary and secondary outcomes from the RCT. Chapter 5 extends those results by comparing the two treatment models on the therapeutic alliance. The majority of the content in this chapter was written by Maria O’Callaghan under the supervision of Dr Carla Walton (PhD candidate) and Dr Sean Halpin. Dr Carla Walton developed the research design. Maria O’Callaghan served as a research assistant on the RCT and used the data from the trial to explore the concept of therapeutic alliance. Ms O’Callaghan submitted a written thesis in partial fulfilment of the requirements for the degree of Master of Clinical Psychology, University of Newcastle. Aspects of that thesis have been adapted and are included here with permission of Ms O’Callaghan and Dr Halpin. 3 January 2018 Ms Maria O’Callaghan Date 3 January 2018 Dr Sean Halpin Date 188 The impact of Borderline Personality Disorder (BPD) on individuals is substantial. People with this condition experience intense emotional pain and distress in their lives (O’Connell & Dowling, 2014). They are high consumers of emergency, inpatient, and outpatient psychiatric services, with up to 84% attempting suicide and 10% dying by suicide (Black et al., 2004). Persons with a diagnosis of BPD experience intense difficulties with interpersonal relationships (O’Connell & Dowling, 2014) and consequently, developing a therapeutic relationship can be particularly challenging. Higher treatment attrition rates are reported for this client group compared to other mental health disorders (Marlowe, Kirby, Festinger, Husband, & Platt, 1997; Martinez- Raga, Marshall, Keaney, Ball, & Strang, 2002). The strength of the therapeutic alliance is reported as the strongest predictor of dropout rates among persons with BPD (Spinhoven, Giesen-Bloo, Van Dyck, Kooiman, & Arntz, 2007; Wnuk et al., 2013). Attrition remains an ongoing concern in the treatment of BPD (Wnuk et al., 2013). Therapists treating BPD need to be vigilant about developing and maintaining robust alliances with this client group to facilitate effective interventions (Gunderson & Links, 2008; Linehan, 1993a). The relationship between therapeutic alliance and outcome is one of the most widely studied topics in the general psychotherapy literature (Horvath, Del Re, Fluckiger, & Symonds, 2011; Ribeiro, Ribeiro, Goncalves, Horvath, & Stiles, 2013). A recent meta-analysis reported an aggregate correlation of r = 0.275 (Horvath et al., 2011). The alliance-outcome relationship has been reported across diverse research conditions, numerous therapies, and various outcome variables (Fluckiger, Del Re, Wampold, Symonds, & Horvath, 2012) and alliance has consistently been shown to be a reliable predictor of outcome, despite only accounting for a small amount of the 189 variance. Consequently, the therapeutic alliance has been firmly established as a significant contributor to therapeutic outcome, and therapists are advised to focus on developing strong alliances with their clients from the commencement of therapy (Horvath & Bedi, 2002). While alliance-outcome research is extensive, there is an incomplete understanding of how the alliance works in therapy and if the alliance works differently in different therapies (Bedics, Atkins, Harned, & Linehan, 2015). A limitation of alliance research to date is the focus on global alliance ratings, without exploring how components of the alliance may affect outcomes differently (Bedics et al., 2015). Bordin (1975) defined three components of the therapeutic alliance (tasks, goals and bond), and predicted that different therapeutic approaches would place different emphasis on these components (Doran, Safran, Waizmann, Bolger, & Muran, 2012). The limited research examining alliance components appears to support Bordin’s predictions. Webb et al. (2011) explored the relationship between alliance components and outcome for depressed participants treated with cognitive therapy (CT). Agreement on tasks and goals significantly related to outcome, whereas the relationship between bond and outcome was not significant. In a structured therapy such as CT, agreement on tasks and goals of therapy may be particularly important, whereas the bond may play a less significant role in symptom change. Webb et al. recommended that future research should replicate this finding and explore the relationship between alliance components and outcome across different mental illnesses, different therapeutic modalities, different methods of assessment, and more than two assessment points. Ulvenes et al. (2012) explored bond and outcome in short-term dynamic psychotherapy compared with CT for clients diagnosed with 190 Cluster C Personality Disorders (Avoidant, Dependent, and Obsessive-Compulsive). Therapeutic interventions focusing on affect were negatively associated with bond in both therapies and positively associated with symptom reduction in dynamic psychotherapy, but not in CT. These studies demonstrate the benefit of including alliance component analysis when investigating alliance, providing a more comprehensive picture of aspects of the alliance associated with outcome. Research on alliance-outcome relationships with BPD is less extensive than across the general psychotherapy literature (Barnicot et al., 2012). However, the alliance has been shown to be a significant predictor of outcome, with stronger alliances associated with improved BPD outcomes (Barnicot et al., 2012; Hirsh, Quilty, Bagby, & McMain, 2012; Turner, 2000). The relationship between alliance components and outcome in BPD was explored in a recent study comparing Dialectical Behaviour Therapy (DBT) with Community Treatment by Experts (CTBE; Bedics et al., 2015). The alliance was measured using the California Psychotherapy Alliance Scale (CALPAS; Gaston, 1991), which identifies four alliance components: working strategy consensus, patient commitment, therapist understanding and involvement, and patient working capacity. Alliance components were reported to work differently in DBT and CTBE, exerting unique effects on outcomes. DBT therapists reported greater goal and strategy agreement early in therapy than CTBE. An increase in total alliance scores as measured by the client was associated with reduced self-harm in DBT but not CTBE. Further analysis identified two components of the client-rated alliance (patient commitment and therapist understanding) associated with reduced self-harm in DBT only. Another alliance component (working capacity) was associated with reduced suicide attempts in DBT only. 191 Improving our knowledge of the relationship between alliance components and outcome may assist with BPD treatment. For example, if some alliance components have a stronger relationship with outcome than others, then therapeutic focus could be modified accordingly. DBT is currently considered one of the most effective treatments for BPD (Cristea et al., 2017) and therefore important to include in further BPD alliance research. However, there are also a number of other therapies with evidence for good outcomes in BPD (Stoffers et al., 2012), which should be included in future alliance research. The Conversational Model (CM) is one such therapy demonstrating encouraging results for BPD (Stevenson & Meares, 1992; Stevenson et al., 2005). To our knowledge, CM therapy has not been included in alliance-outcome research to date. DBT and CM are two distinct therapies, with different theoretical foundations, and place different emphases on alliance components. DBT is a structured therapy focusing on learning and developing new skills, through targeting tasks and goals developed by the therapist and client during therapy. CM is a non- directive therapy where the therapist attunes to the client’s conversation and feelings with the aim of facilitating the client’s emotional maturation. The current study aimed to compare the therapeutic alliance in two different therapies, DBT and CM, for the treatment of BPD. We assessed both therapist-rated and client-rated alliance. First, we predicted greater total alliance ratings in DBT than CM, consistent with Bedics et al. (2015). Second, we predicted greater task and goal ratings in DBT than CM, reflecting the emphasis of these components in DBT. Third, we predicted a significant positive relationship between total alliance ratings and therapeutic outcome in both therapies. Finally, we predicted that task and goal ratings would be more strongly associated with therapist outcome in DBT than CM. 192 5.1 Method Chapter 2 in this thesis contains a more detailed description of the methodology of the trial. A brief summary of methodological components is included in this chapter for the reader’s ease. 5.1.1 Participants Participants were those involved in a randomised clinical trial (RCT) described throughout this thesis. The setting was the Centre for Psychotherapy (CFP), which provides a specialist service for the treatment of BPD and/or Eating Disorders. The RCT compared DBT and CM for the treatment of BPD. General medical practitioners, community mental health teams, and private therapists referred the participants. Participants were required to have a diagnosis of BPD in accordance with the Diagnostic and Statistical Manual of Mental Disorders 4th edition (DSM-IV; American Psychiatric Association, 1994); be aged between 18 and 65 years; and have a minimum of three suicide attempts and/or three non-suicidal self-injury (NSSI) incidents during the 12 months prior to recruitment. Exclusion criteria included: developmental disability; disabling organic conditions; antisocial behaviour that presented a significant threat to staff and other clients; acute psychotic illness; inability to speak or read English; living greater than one hour’s drive from Newcastle; substance dependence other than nicotine; and previous DBT or CM treatment. The inclusion of comorbid Axis I and Axis II disorders was permitted, as was medication use. As the CFP is a public health service, participants did not pay for treatment and were reimbursed $20 for travel and expenses related to attending research appointments. Recruitment commenced in January 2007, with a total sample size of 162 reached by April 2013. Participants were randomly allocated to either DBT (n = 81) or 193 CM (n = 81) using randomisation stratified by gender and current prescribed antidepressant medication, as described in Chapter 2. The study was approved by the University of Newcastle Ethics Committee (Reference Number H-2010-1146) and Hunter New England Human Research Ethics Committee (Reference Number 06/12/13/5.11). 5.1.2 Procedure Potential participants attended an initial assessment with a psychotherapist to determine if they met the criteria for the service and their suitability for treatment. This was followed by a diagnostic interview conducted by a consultant psychiatrist. Participants diagnosed with BPD who agreed to undergo treatment then met with a Research Assistant (RA) who provided details of the research study. They were advised that receiving treatment was not contingent on research participation and that therapists would not have access to their completed research assessments. In the event of discontinuing therapy, they were invited to attend ongoing research assessments. Permission was sought to audio record therapy sessions and appropriate consent forms were signed. Assessments of alliance and outcome measures were conducted at baseline, mid-treatment (seven months), and at the termination of therapy (14 months). Baseline assessments were conducted prior to commencement of therapy with the exception of assessments of the alliance, which were conducted after the sixth therapy session. All assessments were arranged and conducted by an RA who remained blind to treatment condition throughout the research period. Participants were assigned to therapists based on therapists’ availability. Participants who failed to attend four 194 consecutive treatment appointments were classified as having dropped out of therapy but were contacted for follow-up assessments. 5.1.3 Therapists and Research Assistants Treatment was delivered by 32 therapists (6 male and 26 female). Disciplines included psychologists, psychiatrists, psychiatric registrars, social workers, mental health nurses, and occupational therapists. Therapists had a minimum of two years’ clinical experience and an average of 9.5 years’ experience working with BPD. Some therapists provided treatment in DBT only (n = 7), some in CM only (n = 13), and some provided treatment in both models (n = 12). The average therapist caseload was five clients. All therapists were extensively trained in DBT and CM prior to commencement of treatment. Treatment integrity was monitored through weekly supervision. Research Assistants were tertiary qualified in psychology or counselling and were trained to conduct clinical assessments. 5.1.4 Treatments Dialectical Behaviour Therapy. DBT is a cognitive behavioural treatment originally developed to treat suicidal individuals (Neacsiu, Rizvi, & Linehan, 2010). DBT focuses on teaching and practising new skills and facilitating the replacement of maladaptive behaviours with more effective behaviours. The core strategy of DBT involves balancing validation with behavioural change (Linehan, 1993a). DBT involves four components: individual psychotherapy; group therapy; telephone coaching, and consultation amongst therapists. Each week participants attended a one-hour individual session with their therapist and a two-and-a-half-hour group-based skills training. Participants had access 195 to telephone coaching as required. DBT was provided for 14 months. At the commencement of therapy, clients and therapists committed to work together on reducing suicidal or self-harming behaviour, therapy-interfering behaviour, and quality of life-interfering behaviour, while focusing on implementing skills learned in therapy. 5.1.4.1 Conversational Model Therapy. The theoretical basis of CM lies in psychodynamic principles and interpersonal and humanist/experiential concepts (Guthrie, 1999). CM was originally developed with inpatients who exhibited symptoms similar to BPD (Stevenson & Meares, 1992). CM is a non-directive therapy where the therapist attunes to the client’s conversation and feelings. The process aims to assist clients to reflect on their feelings and behaviour and to generalise this reflective capacity in other areas of their lives. CM aims to facilitate the client’s emotional maturation and help them to build a consistent and coherent sense of self. Participants attended individual therapy twice weekly over 14 months with each session being approximately 60 minutes. 5.1.5 Measures Working alliance and trial outcomes were measured using the following instruments that are described in more detail in Chapter 2; the outcome domain is identified first, followed by the measure used. (a) Working Alliance: Working Alliance Inventory (Horvath, Greenberg, & Pinsoff, 1986) (b) Depression severity: Beck Depression Inventory II (BDI-II; Beck et al., 1996). (c) Suicidal behaviour and NSSI: The Suicide Attempt and Self-Injury Count (SASI-Count; Linehan & Comtois, 1996). 196 (d) BPD Severity: Borderline Personality Disorder Severity Index (BPDSI-IV; Arntz et al., 2003) (e) Dissociation: Dissociative Experiences Scale (DES; Bernstein & Putman, 1986) (f) Interpersonal Problems: Inventory of Interpersonal Problems (IIP; Horowitz, Rosenberg, Baer, Ureno, & Villasenor, 1988) 5.1.6 Data Analysis Linear Mixed Modelling (LMM; Brown & Prescott, 2006) was the primary data analytic procedure used on the intent-to-treat sample. LMM is a recommended method for examining longitudinal data, allowing for the ability to account for missing data under the Missing at Random assumption (MAR) and the assessment of time- varying covariates (alliance in this study) and time invariant covariates (Brown & Prescott, 2006). The assumption of normality was checked by plotting histograms of model residuals and was found to be satisfactory for all variables except suicide attempts and self-harm which were moderately right skewed. LMM with robust estimators (which assists in the estimation of significance in the presence of failure of model assumptions) was conducted for these two variables. Appropriate residual covariance structures were used to model correlation due to repeated time measurements and any departures from homoscedasticity. Akaike Information Criteria (Burnham & Anderson, 2004) were used to choose the most appropriate covariance structure for each LMM analysis. SPSS 23 software was used to perform all analyses. 197 Therapeutic alliance was examined using the following model to study alliance change over the course of therapy: time (modelled categorically, levels six weeks, seven months, and 14 months) and treatment (DBT and CM) and time*treatment interaction. Eight models were examined, one for each of the following aspects of the alliance: total alliance and components tasks, goals, and bond as rated by both clients and therapists. Due to multiple models for different alliance measures, a Bonferroni adjustment was applied to control the family wise error rate, with significance level = .05/(4*2) = .006 BPD outcomes across both treatments were examined using the following model: time (modelled categorically, baseline, seven months and 14 months) and treatment (DBT and CM) and time*treatment interaction and alliance (WAI), to study BPD symptom change over the course of therapy. Six models were examined, one for each of the following BPD symptom outcomes: depression severity, dissociative experiences, interpersonal problems, self-harm, suicide attempts and BPD severity index. Bonferroni adjustment was applied to the significance level of the alliance main effect term to control the family wise error rate, with significance level = .05/(6*4*2) = .001. Two outliers were identified at initial assessment, one in CM reporting 152 lifetime suicide attempts and one in DBT reporting lifetime 200 suicide attempts. Analyses were conducted with and without these outlying cases with identical results. Results are presented without these outlying cases. 198 5.2 Results Descriptive statistics for client- and therapist-rated alliance (total and subscale scores) and BPD outcomes are provided in Table 5.1. Higher scores on alliance measures indicate a stronger alliance. Higher scores on BPD outcome measures indicate higher levels of psychopathology. For the most part, ratings of working alliance increased slightly over assessment periods. However, there were marked improvements in BPD outcomes at mid-assessment and on completion of therapy compared to baseline. Specific findings are addressed comprehensively below. Internal consistency in the current study for the total client-rated WAI was excellent (α = 0.95, 0.96, 0.94 at initial, mid- and on completion of therapy), with subscales over the three periods ranging from a minimum value of α = 0.79 (tasks at baseline) to a maximum of α = 0.93 (tasks at mid-assessment). Internal consistency for the total therapist-rated WAI was also excellent (α = 0.94, 0.96, 0.95), with scales ranging from α = 0.69 (bond at initial) to α = 0.92 (goals at mid- assessment). Internal consistency was excellent across all other measures used in the current study: BDI-II α = .96; BPDSI-IV α = 0.84; DES α = 0.98 and IIP α = 0.96. 199 Table 5.1 Descriptive Statistics for Client and Therapist Alliance Rating (WAI) and BPD Outcomes by Assessment Period Baseline Mid Post DBT CM DBT CM DBT CM ALLIANCE (WAI) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Client Total 202.2 (39.0) 193.7 (29.9) 207.8 (33.2) 200.4 (28.8) 216.9 (29.2) 211.5 (25.9) Therapist Total 195.1 (21.5) 181.9 (22.4) 195.7 (22.8) 196.5 (22.1) 201.7 (21.2) 193.1 (28.2) Client Tasks 68.9 (12.2) 64.1 (11.3) 69.5 (11.9) 67.1 (11.6) 74.1 (8.5) 70.4 (10.1) Therapist Tasks 64.6 (8.2) 60.9 (8.3) 63.9 (8.3) 64.0 (9.7) 66.2 (8.1) 62.6 (11.2) Client Goals 68.5 (12.2) 63.0 (11.1) 70.3 (10.8) 65.0 (9.3) 73.9 (7.8) 69.0 (11.4) Therapist Goals 65.0 (8.8) 57.9 (9.3) 64.0 (9.3) 62.5 (9.7) 66.8 (7.7) 61.0 (11.9) Client Bond 64.8 (12.8) 66.6 (10.4) 68.1 (12.9) 68.3 (10.3) 68.9 (15.0) 72.1 (8.5) Therapist Bond 65.5 (6.4) 63.2 (7.4) 67.8 (6.7) 70.0 (5.8) 68.8 (6.9) 69.5 (7.4) BPD OUTCOMES 200 BDI-II 38.6 (10.3) 35.6 (10.7) 22.3 (11.6) 27.2 (13.8) 16.3 (14.5) 21.1 (17.6) DES 27.4 (15.8) 32.5 (19.7) 24.5 (14.3) 29.4 (20.4) 17.3 (12.5) 22.6 (20.1) IIP 126.6 (27.5) 119.8 (32.0) 113.8 (34.9) 117.1 (32.0) 94.3 (45.0) 97.1 (39.5) NSSI 63.2 (111.9) 80.4 (118.6) 26.4 (59.2) 35.9 (64.0) 10.1 (35.9) 15.1 (30.9) Suicide Attempt 6.7 (17.6) 3.7 (6.7) 1.8 (6.9) 1.5 (3.5) 0.7 (2.1) 1.5 (11.2) BPDSI 39.4 (9.8) 39.0 (14) - - 18.0 (9.4) 18.8 (12) Notes. WAI = Working Alliance Inventory; BPD = Borderline Personality Disorder; DBT = Dialectical Behaviour Therapy; CM = Conversational Model Therapy; BDI-II = Beck Depression Inventory II; DES = Dissociative Experiences Scale; IIP = Inventory of Interpersonal Problems; Suic Attem = Suicide Attempts; BPDSI = Borderline Personality Disorder Severity Index. - not assessed at mid-therapy Results for treatment differences in alliance (total and subscale scores) as rated by clients and therapists are provided in Table 5.2. Our primary interest was in time by treatment model interaction. Significant features of treatment model and time effects as seen in Table 5.2 are discussed in more detail below in terms of individual outcomes for therapist and client total, task, goals and bond components. 201 Table 5.2 Linear Mixed Model Results for Client and Therapist Alliance (WAI)—Significance of Effects Rater Alliance Time Treatment Time* (WAI) Model Treatment Model Client Total <.001* .057 .338 Client Tasks <.001* .024 .821 Client Goals <.001* .001* .452 Client Bond <.001* .844 .047 Therapist Total .011 .029 .080 Therapist Tasks .115 .087 .101 Therapist Goals <.001* .001* <.001* Therapist Bond <.001* .349 .070 Note. WAI = Working Alliance Inventory * p < .005; 5.2.1 Client-Rated Alliance (WAI) Figure 5.1 shows the mean client-rated, total alliance over time for both treatments, for which the time by treatment interaction was not significant. Figures 5.2, 5.3, and 5.4 show mean, client-rated alliance components (tasks, goals, and bond) over time for both treatments. Again, the time by treatment interaction for all client- rated alliance components was not significant. 220 C) - 210 200 190 DBT 180 CM 170 Total Alliance (WAI Initial Mid Post Time Figure 5.1. Mean Client-Rated Total Alliance (WAI-C) Over Time 202 75 70 65 C Tasks - DBT 60 WAI CM 55 Initial Mid Post Time Figure 5.2. Mean Client-Rated Tasks Over Time 75 70 65 C Goals - DBT WAI 60 CM 55 Initial Mid Post Time Figure 5.3. Mean Client-Rated Goals Over Time 75 70 65 C C Bond - DBT WAI 60 CM 55 Initial Mid Post Time Figure 5.4. Mean Client-Rated Bond Over Time 203 In the following sections, differences in means are compared and the effect size (the difference between a pair of time periods or treatment models) is indicated by the symbol Md. Figure 5.5 shows the mean client-rated total alliance over time regardless of treatment condition. Client-rated, total alliance increased over time with the increase from baseline to post-assessment being significant Md = 15.23, 95% CI [9.09, 21.37], t(315) = 4.88, p < .001, and increase from mid- to post-assessment being significant Md = 10.68, 95% CI [4.66, 16.69], t(315) = 3.49, p =.001. The increase in client-rated total alliance between baseline and mid- was not significant. Figure 5.6 shows mean client- rated alliance components tasks, goals, and bond over time regardless of treatment condition. Each client-rated alliance component increased over time with the increase between baseline and post- being significant for tasks Md = 5.57, 95% CI [3.26, 7.88], t(318) = 4.75, p < .001, goals Md = 5.64, 95% CI [3.03, 8.24], t(317) = 4.26, p < .001, and bond Md = 4.04, 95% CI [2.02, 6.06], t(316) = 3.94, p < .001. The increase in client- rated alliance components between mid- and post- was significant for tasks Md = 4.09, 95% CI [1.81, 6.36], t(318) = 3.53, p < .001, and goals Md = 4.05, 95% CI [1.90, 6.20], t(317) = 3.71, p < .001, and approached significance for bond Md = 2.57, 95% CI [0.59, 4.55], t(316) = 2.55, p = .011. The increase in client-rated alliance components tasks, goals and bond between baseline and mid-assessment was not significant. 204 220 C) - 210 200 190 180 Total Alliance (WAI 170 Initial Mid Post Time Figure 5.5. Client-Rated Total Alliance (WAI-C) Over Time 75 70 C) - Tasks 65 Goals 60 Bond Alliance (WAI 55 Initial Mid Post Time Figure 5.6. Client-Rated Alliance Components (Tasks, Goals, and Bond) Over Time Client-rated alliance component goals were significantly greater in DBT than CM. The difference between means averaged over all time periods for client-rated goals was Md = 5.77, 95% CI [2.49, 9.05], t(317) = 3.46, p = .001. Client-rated total alliance and components tasks and bond were not significantly different between treatments. 5.2.2 Therapist-rated Alliance (WAI) Figure 5.7 shows the mean therapist-rated, total alliance over time for both treatments, for which the time by treatment interaction was not significant. Figures 5.8, 5.9, and 5.10 show mean, therapist-rated alliance components (tasks, goals and 205 bond) over time for both treatments. There was a significant time by treatment interaction for therapist-rated goals. There was no significant difference between baseline and mid- goals Md = 0.74, 95% CI [-2.21, 3.70], t(321) = 0.50, p = .621 for DBT therapists. From mid- to post-assessment goals increased a small but significant amount Md = 2.67, 95% CI [1.81, 3.52], t(321) = 6.14, p < .001. The pattern was reversed for CM therapists. The increase from baseline to mid- goals approached significance Md = 2.74, 95% CI [-0.08, 5.56], t(321) = 1.91, p = .057, and between mid- and post- there was no significant difference Md = 0.03, 95% CI [-0.82, 0.88], t(321) = 0.07, p = .946. Furthermore, CM therapist goals were lower than DBT therapist goals at each assessment with the difference being significant at baseline assessment Md = 6.49, 95% CI [3.13, 9.84], t(321) = 3.81, p < 001. The time by treatment interaction for therapist-rated alliance components tasks and bond was not significant, indicating no difference between treatments for these therapist-rated alliance components. 206 220 DBT T) - 210 CM 200 190 180 Total Alliance (WAI 170 Initial Mid Post Time Figure 5.7. Mean Therapist-Rated Total Alliance (WAI-T) Over Time 75 DBT 70 CM T Tasks 65 - WAI 60 55 Initial Mid Post Time Figure 5.8. Mean Therapist-Rated Tasks Over Time 75 DBT 70 CM T Goals 65 - WAI 60 55 Initial Mid Post Time Figure 5.9. Mean Therapist-Rated Goals Over Time 207 75 DBT CM 70 T Bond T 65 - WAI 60 55 Initial Mid Post Time Figure 5.10. Mean Therapist-Rated Bond Over Time Figure 5.11 shows the mean therapist-rated total alliance over time regardless of treatment condition. Therapist-rated total alliance increased over time with the difference between baseline and post- being significant Md = 9.13, 95% CI [3.03, 15.22], t(318) = 2.95, p = .003. The increase in therapist-rated total alliance between baseline and mid, and mid and post-assessment was not significant at p = .006 level. Figure 5.12 shows mean therapist-rated alliance components tasks, goals, and bond over time regardless of treatment condition. Therapist-rated goals increased significantly from mid to post-assessment Md = 1.35, 95% CI [0.74, 1.95], t(321) = 4.39, p < .001. Increases in therapist-rated goals between baseline and mid, and baseline and post- were not significant at p = .006. Therapist-rated bond increased significantly from baseline to mid Md = 3.33, 95% CI [1.83, 4.83], t(321) = 4.37, p < .001, and baseline to post-assessment Md = 4.80, 95% CI [2.92, 6.68], t(321) = 5.01, p < .001. The increase in therapist-rated bond from mid to post-assessment was not significant. There were no significant differences for therapist-rated task between any assessment periods. 208 220 T) - 210 200 190 180 Total Alliance (WAI 170 Initial Mid Post Time Figure 5.11. Therapist-Rated Total Alliance (WAI-T) Over Time 70 T) - 65 60 Tasks Goals Alliance (WAI 55 Bond 50 Initial Mid Post Time Figure 5.12. Therapist-Rated Alliance Components (Tasks, Goals, and Bond) Over Time Therapist-rated goals were significantly greater in DBT than CM. The difference between means averaged over all time periods for therapist-rated goals was Md = 5.04, 95% CI [2.13, 7.95], t(321) = 3.41, p = .001. Therapist-rated total alliance and components tasks and bond were not significantly different between treatments at p = .006 level. 5.2.3 Client-rated Alliance and BPD Outcomes Table 5.3 shows Linear Mixed Model (LMM) results for client-rated alliance and BPD treatment outcomes. 209 Table 5.3 Linear Mixed Model Results for Client-Rated Alliance (WAI-C) and BPD Treatment Outcome—Significance of Effects BPD Alliance Time Treatment Time* Alliance Alliance* Treatment Component Model Treatment (WAI-C) Treatment Outcomes Model Model BDI-II Total <.001* .429 .003 .003 .456 BDI-II Tasks <.001* .522 .005 <.001* .241 BDI-II Goals <.001* .675 .006 <.001* .615 BDI-II Bond <.001* .275 .004 .481 DES Total <.001* .109 .843 .008 .933 DES Tasks <.001* .098 .913 .054 DES Goals <.001* .167 .863 .005 .778 DES Bond <.001* .054 .865 .011 IIP Total <.001* .800 .115 <.001* .337 IIP Tasks <.001* .842 .120 .004 .236 IIP Goals <.001* .622 .124 <.001* .596 IIP Bond <.001* .840 .156 <.001* .216 NSSI Total <.001* .376 .192 .228 NSSI Tasks <.001* .431 .309 .391 NSSI Goals <.001* .480 .292 .210 NSSI Bond <.001* .320 .185 .239 Suic Attem Total .014 .720 .399 .864 Suic Attem Tasks .016 .685 .395 .542 Suic Attem Goals .016 .701 .403 .731 Suic Attem Bond .012 .718 .407 .591 BPDSI Total <.001* .984 .914 .235 BPDSI Tasks <.001* .970 .823 .584 BPDSI Goals <.001* .888 .873 .088 BPDSI Bond <.001* .915 .973 .270 Note. WAI = Working Alliance Inventory; BPD = Borderline Personality Disorder; BDI-II = Beck Depression Inventory II; DES = Dissociative Experiences Scale; IIP = Inventory of Interpersonal Problems; Suic Attem = Suicide Attempts; BPDSI = Borderline Personality Disorder Severity Index. * significant at α = .001 level There was a significant negative relationship between client-rated tasks and depression severity b = -0.21, SE = 0.06, t = -3.60, p < .001, CI [-0.33, -0.10], and client- rated goals and depression severity b = -0.23, SE = 0.06, t = -3.77, p < .001, CI [-0.35, - 0.11]. These results indicate that clients with greater total alliance and greater tasks and goals had lower depression scores. The relationship between client-rated bond and depression severity was not significant. There was a significant negative relationship between client-rated total alliance and interpersonal problems (IIP) b = -0.22, SE = 0.06, t = -3.62, p < .001, CI [-0.35, - 0.10], between client-rated goals and interpersonal problems b = -0.59, SE = 0.17, t = - 210 3.55, p < .001, CI [-0.92, -0.26], and between client-rated bond and interpersonal problems b = -0.61, SE = 0.18, t = -3.44, p < .001, CI [-0.96, -0.26]. The relationship between client-rated tasks and interpersonal problems approached significance b = - 0.47, SE = 0.16, t = -2.89, p = .004, CI [-0.79, -0.15]. These results indicate that clients with greater total alliance and greater alliance components rated interpersonal problems lower. The relationships between other BPD treatment outcomes self-harm, suicide and BPD severity index with alliance was not significant. Furthermore, there was no interaction between client-rated alliance and treatment models. Time, treatment model, and time by treatment model interaction results are discussed in Chapter 4 and hence not discussed here. 5.2.4 Therapist-rated Alliance and BPD Outcomes Table 5.4 shows Linear Mixed Model (LMM) results for therapist-rated alliance and BPD treatment outcomes. 211 Table 5.4 Linear Mixed Model Results for Therapist-Rated Alliance (WAI-T) and BPD Treatment Outcomes— Significance of Effects BPD Alliance Time Treatment Time* Alliance Treatment Component Model Treatment (WAI-T) Outcomes Model BDI-II Total <.001* .464 .003 .073 BDI-II Tasks <.001* .440 .003 .046 BDI-II Goals <.001* .535 .003 .082 BDI-II Bond <.001* .403 .004 .424 DES Total <.001* .095 .810 .074 DES Tasks <.001* .116 .902 .060 DES Goals <.001* .130 .857 .146 DES Bond <.001* .092 .842 .260 IIP Total <.001* .732 .096 .047 IIP Tasks <.001* .777 .110 .058 IIP Goals <.001* .730 .194 .121 IIP Bond <.001* .837 .177 .098 NSSI Total <.001* .193 .236 .488 NSSI Tasks <.001* .195 .243 .299 NSSI Goals <.001* .220 .259 .381 NSSI Bond <.001* .181 .236 .955 Suic Attem Total .029 .972 .612 .074 Suic Attem Tasks .019 .986 .628 .084 Suic Attem Goals .023 .828 .603 .045 Suic Attem Bond .036 .926 .595 .274 BPDSI Total <.001* .956 .984 .736 BPDSI Tasks <.001* .960 .709 .364 BPDSI Goals <.001* .933 .943 .776 BPDSI Bond <.001* .935 .883 .327 Note. WAI = Working Alliance Inventory; BPD = Borderline Personality Disorder; BDI-II = Becks Depression Inventory II; DES = Dissociative Experiences Scale; IIP = Inventory of Interpersonal Problems; Suic Attem = Suicide Attempts; BPDSI = Borderline Personality Disorder Severity Index. * significant at α = .001 level There were no significant effects for therapist-rated total alliance or any therapist-rated alliance components with any BPD treatment outcomes. Time, treatment model, and time by treatment model interaction results are discussed in Chapter 4 and hence not discussed here. Therapist and client alliance (total and subscale score) correlations are provided in Table 5.5. All correlations were significant at p = .01. In general correlations were lowest at baseline therapy, highest at mid- therapy and slightly lower at post- therapy than mid-therapy. 212 Table 5.5 Client and Therapist Alliance Correlations Baseline Mid Post Total .404* .581* .517* Tasks .413* .596* .528* Goals .426* .554* .477* Bond .239* .478* .431* * significant at α = .01 level 5.3 Discussion The aim of the current study was to compare the therapeutic alliance in two different therapies, DBT and CM for the treatment of BPD. We predicted that there would be a greater total alliance in DBT than CM, and that the alliance components of tasks and goals would be greater in DBT than CM. However, our results showed no significant difference in client-rated total alliance or client-rated tasks, goals or bond between the two treatments. For client-rated goals, these were significantly greater in DBT than CM. Similar to client-rated alliance results, there was no significant difference in therapist-rated total alliance or therapist-rated tasks and bond between the two treatments. There was a significant time by treatment interaction for therapist-rated goals, which were significantly greater in DBT than CM over time. With the exception of goals, there were no other treatment effects for therapist-rated alliance components. Overall, our finding of no significant difference in total alliance between DBT and CM is consistent with previous research comparing alliance-outcome across different treatments (Horvath & Symonds, 1991). In a meta-analysis of 27 alliance studies Horvath and Symonds found few differences in between treatments for alliance and outcome. They proposed that such homogeneity of results may have stemmed from the use of different alliance measures, with researchers selecting a 213 measure which best captured the theoretical framework being assessed. In our study we compared the alliance in two very different therapies using the WAI, which may be considered more applicable to DBT in terms of its focus on tasks and goals. Despite this, we found no significant difference in total alliance between the two therapies. Our results share some similarities with alliance research conducted by Bedics et al. (2015) who compared alliance in BPD clients treated with DBT and Community Treatment by Experts (CTBE). Bedics et al. reported greater therapist-rated alliance in DBT than CTBE and, similar to our study, the difference over time was not significant. They also reported a treatment by time interaction for therapist-rated working strategy consensus (which measures mutual agreement on goals and general procedures in therapy, and is comparable to goals as measured by the WAI), where DBT therapists reported stronger working strategy consensus early in treatment than in CTBE. In our study, therapist-rated goals were significantly greater in DBT than CM. Furthermore, the pattern of therapist-rated goals over time was very different between the two therapies. For DBT therapists, there was no significant difference between baseline and mid- goals with a small but significant increase between mid- and post- goals. For CM therapists, the pattern was reversed, with an increase in goals from baseline to mid- approaching significance, and no significant difference between mid- and post- goals. Overall, this pattern reflects the emphasis on goals in DBT, which commences early in treatment and is consistent throughout therapy (Linehan, 1993a). In CM there is less focus on goals initially in therapy and it takes longer for goals to become established than in DBT. The difference in patterns of alliance provide support for Bordin (1979) who predicted that different therapeutic approaches would place different emphasis on 214 alliance components. Webb et al. (2011) also hypothesised that in structured therapies agreement on tasks and goals may be particularly important. While we found no significant effects for tasks we found both therapist- and client-rated goals were significantly greater in DBT than CM, highlighting the importance of goals in DBT. In the current study, client-rated total alliance (regardless of treatment condition) increased significantly over time with differences between baseline and post, and mid- to post-assessments being significant. Client-rated tasks, goals, and bond showed a similar pattern of increase over time. Therapist-rated total alliance (regardless of treatment condition) also increased steadily over time with a significant increase between baseline and post-assessment. This trajectory of alliance development in our research is different to that reported in general alliance literature where alliance has been described as developing early in treatment and not changing significantly over time (Horvath, 2000). It is possible that alliance development is more protracted for clients with BPD. Features of BPD such as interpersonal sensitivity and emotional lability have been associated with poorer alliance and treatment attrition (Wnuk et al., 2013). An extended pattern of alliance development was also reported by Spinhoven et al. (2007) in their research with BPD clients utilising different treatment conditions. Our other prediction proposed that there would be a significant relationship between total alliance and therapeutic outcomes (regardless of treatment condition). Furthermore, we predicted that tasks and goals would be more strongly associated with therapeutic outcomes in DBT than CM. The following BPD treatment outcomes were assessed: depression severity; dissociative experiences; interpersonal problems; NSSI; suicide attempts; and BPD severity. We found a significant negative relationship 215 between client–rated tasks and goals with depression severity, and the relationship between the client-rated total alliance with depression severity approached significance. These results indicate that clients who rated tasks and goals higher, had lower depression scores. Furthermore, there was a trend where clients who rated total alliance higher, had lower depression severity ratings. Therefore, it appears that a good alliance is important in the treatment of depression or alternatively that lessening depression allows for the possibility of developing a good alliance. These results are compatible with findings by Hirsh et al. (2012) who reported greater alliances associated with lower depression severity. Likewise, Bedics et al. (2015) reported a trend association between an alliance component (Patient Working Capacity) and depression severity, with increased client-rated Patient Working Capacity associated with decreased depression scores in DBT. We found a significant negative relationship between client-rated goals, bond and total alliance with interpersonal problems, indicating that clients who rated goals, bond and total alliance higher had lower interpersonal problems. Furthermore, the relationship between client-rated tasks and interpersonal problems approached significance indicating a trend between higher client-rated tasks and lower interpersonal problems. There were no further significant findings between client- rated alliance and remaining BPD outcomes measured: NSSI; suicide attempts and BPD severity index. This is in contrast to Bedics et al. (2015) who reported a relationship between alliance with both NSSI and suicide attempts. The interaction between client- rated alliance and treatment model was not significant, indicating no significant difference in the relationship between alliance and outcome across treatment models. In contrast to client-rated alliance results, there was no significant relationship 216 between therapist-rated total alliance or any therapist-rated alliance components and BPD treatment outcomes. The current study has added to extensive alliance literature reporting an alliance-outcome relationship (Bedics et al., 2015; Turner, 2000). However, as outlined above the relationship between total alliance and outcome was present only for client- rated alliance scores and only for selected outcome measures: depression severity and interpersonal problems. This raises the question of why alliance was associated with some outcome measures and not others. Alliance is distinct from the therapeutic relationship (Bedics et al., 2015). It may be possible that the interpersonal relationship between therapist and client has a greater effect on other BPD treatment outcomes such as self-harm and suicide. Comparison of client and therapist alliance correlations showed an interesting trend, where correlations were lowest at baseline assessment, highest at mid- assessment and slightly lower at post-assessment than mid-assessment. This indicated that clients and therapists were in strongest agreement regarding their alliance at mid- therapy, with a slight reduction in agreement at post-therapy. At post-therapy the alliance may be less important as clients have completed therapy which may explain the slight reduction in client-therapist correlation. This study has a number of strengths. It was part of a large RCT using an intent- to-treat design in a public mental health setting. Participants were assessed at each assessment regardless of whether they remained in treatment, and alliance was assessed from both therapist and client viewpoints. The study compared the alliance in two different treatments (DBT and CM) in the treatment of BPD and extended alliance research to include alliance components, in addition to looking at total alliance scores. 217 There were also a number of limitations in this study. We did not account for the possibility that early symptom improvement prior to measurement of the alliance at the sixth session may have influenced alliance ratings. However, previous research has demonstrated that the alliance assessed early in therapy is not simply a result of early symptom change, and that alliance continues to predict treatment outcomes even when controlling for early changes (Barber, Connolly, Crits-Christoph, Gladis, & Siqueland, 2009). There is the possibility of a “halo effect” in our data, as clients who form a strong alliance might also optimistically report more favourable outcomes. Assessments were self-report and may have been subject to response and acquiescent biases. In conclusion, this is the first study that we are aware of to compare alliance components in DBT and CM for BPD. Future research should involve replication of the current study as well as to compare alliance components in other therapies reporting positive outcomes for BPD such as schema-focused and transference-focused therapy (Stoffers et al., 2012). Furthermore, it would be interesting to determine where alliance reaches its maximum potential for BPD clients. In our study total alliance increased across therapy with a significant difference between mid- and post-therapy for client-rated alliance, indicating that maximum alliance was achieved somewhere between 7 and 14 months. Future alliance studies, assessing the alliance at more regular intervals could assist in determining where alliance peaks for BPD clients in their therapeutic journey. This would be useful feedback for clinicians and a reminder of the need to continue focusing on the alliance with this group of clients for an extended period of time. Exploring the alliance in different sub-sets within the full sample based on client BPD severity, co-morbid depression or interpersonal problem 218 severity may also provide additional information regarding extended alliance development. 219 Chapter 6 Characterising the Clinical Impact of Dialectical Behaviour Therapy and the Conversational Model Chapters 4 and 5 presented aggregated data of the results of an RCT on primary and secondary outcomes as well as the impact of each therapy on the therapeutic alliance. Chapter 6 takes the focus from average scores to examining improvement and deterioration at the individual level. 220 Borderline Personality Disorder (BPD) is a disabling mental disorder characterised by instability in a range of domains including affect, interpersonal relationships, identity, and behavioural dysregulation (Lieb et al., 2004). It is associated with a high degree of suffering, efforts to deliberately inflict harm to oneself, high rates of suicide attempts, and a lifetime suicide mortality rate of approximately 10% (Black et al., 2004). There is a high rate of co-morbidity with Axis I mental disorders, which has been estimated at 85% (Leichsenring et al., 2011). Beyond high levels of symptomatic impairment, large-scale studies have shown pervasive social and functional impairment (Gunderson et al., 2011). Until the 1990s, there was little hope about the capacity for BPD to be successfully treated (Korner & McLean, 2017). In the past 20 years, there has been considerable progress in psychotherapeutic treatments developed and evaluated for BPD and most have been shown to be more effective than control interventions for BPD-related problems, such as suicidality and non- suicidal self-injury, hopelessness and depression severity (Cristea et al., 2017). When considering differing types of evidence for a treatment, Howard, Moras, Brill, Martinovich, and Lutz (1996) in their seminal paper, state that there are three fundamental questions that can be asked. These questions are: “(a) does it work in special, experimental conditions?, (b) does it work in practice?, and (c) is it working for this patient? (p.1059).” In regards to DBT, question (a) has been largely answered in previous published research (see Chapter 1) in RCTs and has been shown to work in special, experimental conditions. As discussed in Chapter 1, Conversational Model (CM) has been evaluated in studies with less scientific rigour, and not as part of a randomised trial under well- 221 regulated experimental conditions. Question (b) speaks to treatment effectiveness that asks whether the treatment produces beneficial results when administered in actual clinical settings. In relation to CM and DBT, this was the focus of Chapters 4 and 5. In a routine clinical setting, both therapies were associated with significant improvements in a large range of domains with 14 months of therapy. On average, patients improved. Both questions (a) and (b) seek answers about the general impact of particular interventions and are answered in terms of the average response for each treatment group. Most clinicians are interested in the answer to question (c) of whether a treatment with demonstrated efficacy will work for an individual patient. This question is answered by examining each patient’s response to the treatment provided by a particular clinician. To date, in the field of BPD research, the question of whether or not individual patients have responded to the treatment they received has the least systematic research despite being of the most immediate, day to day concern to clinicians. Every patient will not have the same outcome, even to a well-standardised treatment for a well-specified clinical problem (Howard et al., 1996). Further, questions (a) and (b) are answered using statistically significant differences, rather than whether a patient has obtained a clinically significant change. While a well- designed study is powered for a certain magnitude of change in aggregate difference, it is possible for the results of a study to be statistically significant, but too small to make a difference to the consumer. The perspective of clinical significance previously received less emphasis (Man‐Son‐Hing et al., 2002) and the impact for individuals rather than the group effect is still infrequently discussed. Jacobson and Truax (1991) introduced the concept of clinically significant change to identify when a patient shifts 222 from the dysfunctional range to a functional range during the course of therapy (where the functional range is most commonly defined as extending to two standard deviations above the mean in the direction of functionality for that population for a continuous outcome). Only a small number of the RCTs completed in the area of BPD have reported clinically significant change. Bohus et al. (2004) examined DBT delivered in an inpatient setting. Apart from reporting on statistical significance, they also reported on clinically relevant change, following the suggestions of Jacobson et al. (1999) and selected a general symptom checklist (SCL-90-R) as the main outcome variable. They reported that 13 of the 31 patients in the DBT treatment arm, comprising 41.9% of the sample could be considered “recovered” according to the estimate of clinically significant change. Bos et al. (2010) also used the SCL-90 to evaluate clinically reported change and identified that at post treatment, 58% of patients could be considered recovered. It needs to be considered which domains we define as constituting recovery for people with BPD. The SCL-90-R is a generic measure of general symptoms, frequently used in mental health research, but it is unclear what it tells us about specific recovery in BPD. Analysis of clinical significance requires comparison with community and psychiatric norms and for many of the measures used in the BPD literature, these do not exist. Gratz (2006; 2014) in both of her studies of ERGT calculated the clinical significance of the treatment effect in accordance with Jacobson and Truax (1991). In the 2006 study they found that 50% of participants in the ERGT group reported a reliable improvement from pre- to post-treatment for BPD symptoms, 50% for depressive symptoms and 35% for anxiety symptoms. Those proportions were similar for BPD symptoms and anxiety symptoms in the follow-up study in 2014, but a lesser 223 proportion (30%) for depressive symptoms. Koons et al (2001) reported that in their small sample of DBT patients (n=10), on 6 out of 7 variables, majority of DBT patients (60-80%) met the criterion for clinically significant change, although it was not stated how this was calculated. While the concept of clinically significant change allows us to identify change that is meaningful, one of the criticisms of the concept was noted by Kazdin (1999) who identified that clinically significant change can occur when there is “a large change in symptoms, a medium change in symptoms and no change in symptoms” (p.332). Kazdin suggested that a continuum may be more appropriate, rather than a fixed cut- off. In a later paper, Kazdin (2001) further argued that there was no existing evidence to show that passing a threshold or entering a range corresponds with the client being “better in any way that affects daily functioning or that a failure to pass this threshold means otherwise” (p.461). Jacobson and colleagues (1984) acknowledged that “it is not always possible to identify an appropriate normative group” (p.342) and that in some circumstances the “return to normal” criteria may be too stringent. Wise (2004) reports that many previous articles about clinical significance suggest that the “return to normal” criterion should be the definition of clinical significance. However, Wise also suggests that this may be unrealistic for a range of complex clinical populations, including those with personality disorders, comorbid mental and substance disorders or comorbid mental and medical disorders. Wise notes that studies of outpatients with high prevalence disorders, such as depression, indicate that few achieve complete remission. It is worth considering what would be considered clinical improvement for BPD. At a simplistic level, we could consider someone recovered if they no longer meet 224 diagnostic criteria for the disorder. Very few studies in the area of BPD have reported whether participants still met diagnostic criteria at the end of treatment. Doering et al. (2010) reported that after a year of TFP, 42% of patients no longer met diagnostic criteria for BPD (compared with 15% in the TAU group). Giesen-Bloo et al. (2006) found that approximately 40% of participants that received SFP were recovered (based on scores on the BPDSI), compared with 20% of participants that received TFP. In a similar pattern, the original study evaluating CM (Stevenson & Meares, 1992) reported that 30% of patients no longer met the DSM criteria for BPD at the end of 12 months of treatment (Stevenson et al., 2005). Jorgensen et al. (2012) in their study of MBT and supportive therapy, found that at post-treatment, 52% of patients in the active treatment group no longer met diagnostic criteria for BPD. It is noted that their post- treatment assessment point was two years rather than the other studies described which assessed this at one year. Alongside the aforementioned metrics of recovery, we need to consider what other ways that we might define recovery in BPD. Traditional notions of recovery in mental health have focused on remission of symptoms or no longer meeting diagnostic criteria as well as return to previous levels of functioning (Davidson & Roe, 2007; Slade, 2009; Le Boutillier et al, 2015). For BPD, part of the DSM diagnostic criteria is that the dysfunction is “pervasive.” Hence, it is likely that there was not a previous time without symptoms or substantially better functioning to return to. Further, in DSM, a minimum of five out of nine criteria are required to receive the diagnosis of BPD, so a person could move from five criteria to four criteria and be considered to be ‘recovered’ in terms of no longer meeting the criteria for the disorder, while another 225 person could move from nine criteria to six criteria and be considered ‘not recovered’ by the same standard. It has been noted that there is a dearth of research into the effects of BPD specific interventions beyond symptom change (Bateman et al, 2015; Chanen, 2015; Leichsenring et al., 2011). Consequently, we know little about the impact of these interventions on individuals’ functioning. In qualitative research by Katsakou et al. (2012) and Lariviere et al. (2015), consumers with BPD who had received psychotherapy reported that they did not feel that the word “recovery” accurately captured their experience. They viewed the word “recovery” as implying a dichotomous classification of problems and their perception was that it is used interchangeably with “cured” and that was not their experience. They found it hard to imagine not having some difficulties in dealing with their emotions and their lives and instead preferred the terms: “journey” or “progress” or “learning.” Ng, Bourke, and Grenyer (2016) conducted a systematic review of recovery in BPD from the point of view of consumers, clinicians, family and carers and concluded that instead of symptom remission, a fuller definition of recovery in BPD would include “maintaining sub-threshold symptom expression, engaging in vocational activities that are personally meaningful and sustaining close personal relationships” (p.17). An alternative to using clinically significant change for BPD, that may help to identify the sub-threshold symptom expression discussed above, is the Minimal Clinically Important Difference (MCID; Chan, 2013). MCID considers the smallest amount of improvement that would be considered worthwhile to the patient. MCID may be a better choice than clinically significant change, given it allows for differing 226 baselines and allows for the possibility of improvement without shifting diagnostic categories, which is important in a severe and symptomatic population. MCID has been calculated for a number of outcome measures used to assess improvement in mental health disorders (McMichael et al, 2016). Some have identified the MCID as a change in number of points on a scale, while others identified it as a percentage reduction. In examining MCID for depression, Button et al. (2015) examined the Beck Depression Inventory, 2nd edition (BDI-II; Beck et al., 1996) among a sample of 1039 patients by examining change scores on the BDI-II in relation to a patient’s self-reported global sense of improvement. They found strong evidence that those with a higher BDI-II score before treatment needed a larger change score in order to feel better. As such, they suggested that the MCID is best assessed on a ratio scale that uses percentage of reduction in scores from baseline. This then allows for differences in baseline severity to be taken into account. To date, there has been no published research using MCID for BPD. Hence, there are no established cut-offs identified by experts or patients. As outlined in this introduction, MCID may be a useful way to examine change at an individual level for persons with BPD. This chapter aims to (1) identify change in number of DSM BPD criteria met for individuals between baseline and post-treatment (14 months) and whether participants still met DSM criteria for BPD following 14 months of treatment; (2) identify MCID categories for individuals on number of events of suicide attempts and non-suicidal self-injury (NSSI) and percentage reduction or increase on continuous variables of depression severity, BPD severity, interpersonal problems, mindfulness skills, dissociation, emotion regulation difficulties and sense of self; (3) explore relationships between change in number of BPD criteria met and MCID category on a 227 measure of BPD severity; and (4) compare aggregated scores between the two treatments for the sum of these improved clinical domains. 6.1 Method 6.1.1 Participants Participants were those that had consented to be part of the RCT described in previous chapters. It was necessary to have data from both baseline and the end of treatment (14 months) in order to determine if they had improved or not. Hence, participants were included in the analyses if they had completed 14 months of treatment and completed both the baseline and at least part of the post-treatment assessment (n=114). Some participants did not want to complete all of the assessment and were asked to complete as much as they were willing to. All participants met DSM-IV (American Psychiatric Association, 1994) criteria for BPD and reported a minimum of three episodes of suicide attempts and/or NSSI in the 12 months before recruitment. Exclusion criteria included individuals with a lifetime history of meeting criteria for schizophrenia, bipolar disorder, schizoaffective disorder, or psychotic disorder not otherwise specified. Additional exclusion criteria included intellectual disability, a seizure disorder requiring medication, substance use dependence or mandated court order treatment. Participants had been randomly assigned to either DBT (n=81) or CM (n=81). 51 participants who had completed DBT (63% of those in the intention to treat analysis) and 63 participants who had completed CM (78% of those in the intention to treat analysis), attended both baseline and at least some part of the post-treatment assessment and are included in the data reported in this chapter. 228 6.1.2 Procedure Details on the procedure and further details on the sample can be found in Chapter 2. 6.1.3 Measures Diagnosis of BPD was made by a Consultant Psychiatrist at the baseline assessment using the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (SCID-II; DSM-IV; First et al., 1995) and by a research assistant blind to treatment condition at the post-assessment time point of 14 months (by that time, the consultant psychiatrists who were clinicians on the team, were no longer blind to treatment condition). In addition to the SCID-II, the following clinician administered measures were used (more detail regarding these measures is provided in Chapter 2): (a) Suicidal behaviour and NSSI: The Suicide Attempt and Self-Injury Count (SASI-Count; Linehan & Comtois, 1996). (b) BPD Severity: Borderline Personality Disorder Severity Index (BPDSI-IV; Arntz et al., 2003) The following self-report measures were used: (a) Depression severity: Beck Depression Inventory II (BDI-II; Beck et al., 1996). (b) Dissociation: Dissociative Experiences Scale (DES; Bernstein & Putman, 1986) (c) Interpersonal Problems: Inventory of Interpersonal Problems (IIP; Horowitz, Rosenberg, Baer, Ureno, & Villasenor, 1988) (d) Sense of self: Sense of Self Inventory (SSI; Basten, 2008) 229 (e) Mindfulness: Kentucky Inventory of Mindfulness Skills (KIMS; Baer, Smith, & Allen, 2004) (f) Emotion Regulation: The Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004) 6.1.4 Data Analysis In the absence of any previous literature defining MCID for the measures used in this study, other than for the BDI-II (Button et al., 2015), we followed the recommendation of Button et al. (2015) regarding the importance of use of ratio scales of change rather than specific cut-offs. For the current analysis, we adopted 25% improvement from baseline as the MCID threshold for each of the continuous outcomes. Hence, for all continuous measures, MCID was computed by calculating 25% improvement or 25% deterioration from individual baseline scores and individuals were then classified as improved, unchanged or deteriorated. The participants were all sufficiently severe at baseline assessment on every measure to have scope to improve and as such, all with baseline and post scores were retained in the sample. For measures of suicidal behaviour and NSSI, MCID was calculated by taking an individual’s score on the SASI-Count at baseline (measuring the 7 months prior to the baseline assessment) as well as their score at post (measuring the 7 months prior to the post assessment (14 months) and determining whether the behaviour had increased, stayed the same or decreased. If it had decreased, we looked at whether there had been no episodes of self-injurious behaviour at all in the past seven months or whether there were fewer episodes of the behaviour, but the behaviour had continued. Based on the above information, individuals were then classified into the following categories: ‘deteriorated’; ‘unchanged’; ‘some improvement’ (improved but 230 the behaviour had not fully ceased); or ‘improved’, such that there had been no episodes in the past seven months. A proportion had not reported any self-injurious behaviour at baseline (and hence, had no capacity to improve). Overall, 46 people (40%) had not had a suicide attempt in the seven months prior to the baseline assessment (23 (45%) in DBT condition and 23 (37%) in CM condition). Overall, four people (4%) had not engaged in NSSI in the seven months prior to the baseline assessment (2 (4%) in DBT condition and 2 (3%) in CM condition). The primary focus of analyses was on characterising and describing the sample. Analyses of difference between treatments on MCID categories were calculated using Pearson Chi-Square. An aggregated score was calculated for each individual on the number of outcomes on which it was possible to improve and evaluated the domains discussed above. Deterioration was not calculated as the number of participants who deteriorated (5% averaged across outcomes). The treatment groups were compared in a series of planned t-tests. 6.2 Results 6.2.1 BPD Criteria All participants met DSM-IV diagnostic criteria at baseline assessment (as this was one of the inclusion criteria). Of 114 participants who completed the post- assessment, 43 (38%) no longer met criteria for BPD at post-assessment. Of those with the highest BPD severity on the Borderline Personality Disorder Severity Index, 77 (48%) of the 162 participants who completed the baseline 231 assessment met all nine BPD criteria at baseline. Only 13 (11%) of the 114 participants who completed the post-assessment still met nine criteria at post-treatment. Examining these changes by treatment condition, 21 of the 51 DBT participants no longer met BPD criteria (41%) and 22 of the 63 CM participants no longer met BPD criteria (35%). 6.2.2 Minimal Clinically Important Differences MCID categories for suicide attempts and NSSI are presented in Table 6.1. Across the 68 participants who reported a suicide attempt in the 7 months before the baseline assessment, 50 (74%) reported not having made a suicide attempt in the seven months prior to post-treatment assessment. Despite being randomised, a slightly higher proportion of participants reported suicide attempts prior to treatment in the CM group (63%, compared with 55% of those allocated to DBT). Most (83%) of those in the CM condition at the post-treatment assessment point had improved and ceased suicidal behaviour, i.e., they had not had a suicide attempt in the previous seven months (since the mid-assessment point), and 60% in the DBT condition had ceased the behaviour. Of those who had not ceased suicidal behaviour, 29% in the DBT group and 5% in the CM group had improved and reduced suicidal behaviour. The percentage across both groups who had deteriorated was similar (10% for CM and 11% for DBT). The picture of change for NSSI was slightly different than for suicide attempts. Here, there was a slightly higher number of people who had deteriorated (21% for CM and 12% for DBT) and who had improved in the behaviour but hadn’t ceased it (46% for CM and 57% for DBT). Correspondingly, the numbers who had improved and completely ceased the behaviour were lower (33% for CM and 31% for DBT) than they 232 were for suicide attempts. The differences between CM and DBT were not significant for suicide attempts (Pearson Chi-Square X2 = 8.696, p=0.069) or NSSI (Pearson Chi- Square X2 = 2.053, p=0.562). 233 Table 6.1 Characteristics of Individual Client-Level Outcomes of Minimally Clinically Important Differences (MCID) Analysis for Suicide Attempts and Non-Suicidal Self-Injury (NSSI) Measure MCID category DBT CM Total N (%) N (%) N (%) Suicide Total N 28 (100) 40 (100) 68 (100) Attempts Deteriorated 3 (11) 4 (10) 7 (10) Unchanged 0 (0) 1 (2) 1 (1) Some improvement (didn’t 8 (29) 2 (5) 10 (15) cease behavior) Improved 17 (60) 33 (83) 50 (74) (ceased behavior) NSSI Total N 49 (100) 61 (100) 110 (100) Deteriorated 6 (12) 13 (21) 19 (17) Unchanged 0 (0) 0 (0) 0 (0) Some improvement (didn’t 28 (57) 28 (46) 56 (51) cease behavior) Improved 15 (31) 20 (33) 35 (32) (ceased behavior) Scores are from Suicide Attempt Self Injury Interview count Note. Only those who reported the behaviour at baseline are included in this table. Overall, 46 people (40%) had not had a suicide attempt in the 7 months prior to the baseline assessment (23 (45%) in DBT condition and 23 (37%) in CM condition) and 4 people (4%) had not reported NSSI in the 7 months prior to the baseline assessment (2 (4%) in DBT condition and 2 (3%) in CM condition). Table 6.2 presents the MCID categories for each of the continuous variables. On depression scores (as measured by the BDI-II), BPD severity scores (as measured by the BPDSI), and dissociation scores (as measured by the DES), the majority of participants fell in the improved category with no difference by treatment category (Pearson Chi- Square X2 for depression scores: 4.159, p=0.125; for BPD severity scores: 0.953, p=0.621; for dissociation: 1.699, p=0.428). There was a different pattern for interpersonal problems (as measured by the IIP), sense of self (as measured by the SSI), mindfulness (as measured by the KIMS) and difficulties in emotion regulation (as measured by the DERS) where fewer people were classified as improved. 234 The differences between CM and DBT were not significant for interpersonal problems or sense of self (Pearson Chi-Square X2 for interpersonal problems: 3.897, p=0.142; for sense of self: 2.055, p=0.358). For mindfulness and difficulties in emotion regulation, there were significant differences between the treatments. Greater proportions of the participants in the DBT condition were classified as improved than those in the CM condition on mindfulness (Pearson Chi-Square X2 = 6.731, p<.05), with a similar pattern for difficulties in emotion regulation (Pearson Chi-Square X2 = 17.049, p<.001). Across most outcomes there were a minority of people that were classified as having deteriorated. For sense of self, difficulties in emotion regulation, mindfulness skills and borderline severity scores, the percentage who deteriorated was 3% or less across the whole sample. However, 6% of participants deteriorated in terms of depression scores and 14% of participants deteriorated in terms of dissociation, in both cases, spread between the two treatments. For interpersonal problems, 7% of participants deteriorated across the sample, however, in terms of treatment model, this was primarily among the CM participants (12%) in contrast to 2% for DBT. The Chi- Square analyses were across all categories of improved, unchanged and deteriorated and despite the gap between treatment models on interpersonal problems for deteriorated participants were non-significant across categories (Pearson Chi-Square X2 = 3.897, p=0.142). 235 Table 6.2 Characteristics of Client-Level Outcomes of Minimally Clinically Important Differences (MCID) Analysis for Continuous Variables Measure MCID category DBT CM Total N (%) N (%) N (%) BDI-II Total N 50 (100) 60 (100) 110 (100) Deterioration ≥ 25% 3 (6) 4 (7) 7 (6) Unchanged 7 (14) 18 (30) 25 (23) Improvement ≥ 25% 40 (80) 38 (63) 78 (71) BPDSI Total N 51 (100) 61 (100) 113 (100) Deterioration ≥ 25% 1 (2) 3 (5) 4 (3) Unchanged 9 (18) 13 (21) 22 (20) Improvement ≥ 25% 41 (80) 46 (74) 87 (77) IIP Total N 51 (100) 60 (100) 111 (100) Deterioration ≥ 25% 1 (2) 7 (12) 8 (7) Unchanged 27 (53) 28 (46) 55 (50) Improvement ≥ 25% 23 (45) 25 (42) 48 (43) KIMS * Total N 51 (100) 60 (100) 111 (100) Deterioration ≥ 25% 1 (2) 0 (0) 1 (1) Unchanged 26 (51) 52 (87) 78 (70) Improvement ≥ 25% 24 (47) 8 (13) 32 (29) DES Total N 51 (100) 57 (100) 108 (100) Deterioration ≥ 25% 8 (16) 7 (12) 15 (14) Unchanged 9 (17) 16 (28) 25 (23) Improvement ≥ 25% 34 (67) 34 (60) 68 (63) DERS * Total N 39 (100) 42 (100) 81 (100) Deterioration ≥ 25% 1 (3) 1 (2) 2 (3) Unchanged 13 (33) 26 (62) 39 (48) Improvement ≥ 25% 25 (64) 15 (36) 40 (49) SSI Total N 39 (100) 43 (100) 82 (100) Deterioration ≥ 25% 1 (3) 0 (0) 1 (1) Unchanged 18 (46) 25 (58) 43 (53) Improvement ≥ 25% 20 (51) 18 (42) 38 (46) Key for measures: BDI-II = Beck Depression Inventory II, BPDSI = Borderline Personality Disorder Severity Index, IIP = Inventory of Interpersonal Problems, KIMS = Kentucky Inventory of Mindfulness Skills, DES = Dissociative Experiences Scale, DERS = Difficulties in Emotion Regulation Scale, SSI = Sense of Self Inventory Note. All participants that attended the post-assessment were included in the analysis as based on their baseline scores all had the capacity to improve by 25%. * indicates measure with significant difference between treatment conditions. See text for further information. 236 We examined those who had made a 25% or greater improvement on the BPDSI-IV and what the correspondence was with those who no longer met criteria for BPD. Overall, 87 people out of 113 had improved by 25% or more, which equates to 77% of the sample. However, only 39 of the 87 people (45%) of those who were categorised as improved on the BPDSI no longer met diagnostic criteria for BPD. Figure 6.1 shows the comparative profiles of improvement between the treatment groups. This is a subset of the data from Tables 6.1 and 6.2 that focuses specifically on those who were categorised as improved on the MCID for each outcome in order to compare across outcomes. Data from those who had deteriorated are not shown, as the numbers were small. As can be seen in Figure 6.1, the majority of participants improved in terms of suicide attempts and NSSI (including those who had some improvement (but didn’t cease the behaviour) and those who has improved (and ceased the behaviour completely). The majority also improved in terms of severity of BPD. More than 60% improved their depression scores and more than 50% of participants improved on dissociation, although there was less improvement on these domains than on suicide attempts and NSSI. Less than 50% of participants improved on sense of self and interpersonal problems. 237 KIMS DERS SSI IIP DES CM BDI DBT BPDSI Measure of improvement NSSI Suic Attempt 0 20 40 60 80 100 Percentage of participants who improved Figure 6.1. Profiles of Improvement Between Participants in Different Treatment Conditions Across All Domains 6.2.3 Mean number of outcomes of improvement and deterioration Of the nine outcomes possible to improve on across Tables 6.1 and 6.2, we calculated the mean number of outcomes of improvement and deterioration. Across both treatments, the mean number of outcomes of improvement was 4.75 (SD = 2.43) with a range of 0 to 9. The mean number of outcomes improved on was significantly higher for DBT with a mean of 5.39 (SD = 2.32) compared with CM with a mean of 4.24 (SD = 2.41; F=6.657, df = 112, p<.05). Across both treatments, the mean number of outcomes of deterioration was 0.56 (SD = 0.92) with a range of 0 to 4. The mean number of domains on which deterioration occurred was not significantly different between treatments. Both treatments had a mean number of domains of deterioration less than 1 (DBT = 0.49, CM = 0.62; F=.548, p=ns). 238 6.3 Discussion This chapter examined improvement and deterioration at an individual level as well as between the treatment groups. We found that 38% of participants no longer met criteria for BPD at post-treatment. The majority of participants had improved in terms of number of suicide attempts and non-suicidal self-injury. There was a similar pattern of improvement for BPD severity, depression scores, and dissociation. However, fewer participants improved on the domains of interpersonal problems, sense of self, difficulties in emotion regulation, or mindfulness. Linehan (1993a) suggests that the first target of treatment in DBT is behavioural control, and it would appear that the majority of participants achieved this as shown in the reduction of suicide attempts and NSSI. However, this does not account for the same pattern occurring for CM whereby the majority of participants also improved in numbers of suicide attempts and NSSI. These results may reflect that suicide attempts and NSSI are more responsive to treatment earlier than other outcomes. Fewer participants improved on the domains of interpersonal problems, sense of self, difficulties in emotion regulation and mindfulness. There are a number of potential explanations. It is possible that the gains that occurred are all that is possible with these therapies or alternately, that these outcomes are slower to respond to intervention for some with fewer numbers of participants improving representing greater variability in response. This is consistent with the findings of Korner et al. (2008) who found that participants who received two years of CM rather than one year made greater changes on affective domains. Similarly, Wilks, Korslund, Harned, and Linehan (2016) noted a slow rate of improvement in functional impairment for persons with BPD who had received DBT. 239 Interestingly, while the majority of participants were categorised as improved on a measure of BPD severity, less than half (45%) of those no longer met diagnostic criteria for BPD. Consistent with the literature on recovery in BPD (Ng et al., 2016), this brings into question the possibility of achieving “recovery” according to the diagnostic criteria for BPD. On most of the outcomes, there was a very small percentage of people who deteriorated; ultimately we would want that to be none at all. This could be due to a number of reasons: there may have been ceiling effects with the measures such that there was little room for deterioration; there may have been a response bias from the participants, given that the participants included in these analyses are those for whom we had a post-treatment score and were more likely to have stayed in treatment. The outcome with the highest rate of deterioration was dissociation, with 14% of participants deteriorated, and similar proportions of deterioration across the two treatments. It is established that dissociation is commonly found in persons with BPD (Paret, Hoesterey, Kleindienst, & Schmahl, 2016) and previous research has shown that high levels of dissociation predict a poor response to treatment in BPD (Kleindienst et al., 2011). Hence, this finding highlights that we need better ways to identify people during their treatment who are not responding or are worsening as a result of the treatment. Rizvi (2011) outlined various ways that patients can “fail” in DBT treatment, and it would be worthwhile to conduct case study explorations of the individuals who deteriorated to learn where we might prevent this in future. As was the case with the aggregated results from the RCT presented in Chapter 4, the pattern of findings for the two different treatment conditions was fairly similar. With the findings from the General Linear Model analyses reported in Chapters 4 and 240 5, there were differences in favour of DBT for mindfulness and emotion regulation as well as depression severity, but no differences for the other six outcomes. In the current analyses, the only significant differences between the treatment conditions were for improvements on mindfulness and emotion regulation in favour of DBT. Hence, the current analyses support the findings that DBT is associated with greater improvements in mindfulness and capacity to emotionally regulate than CM. These outcomes are measures of skills deficits and direct treatment targets in DBT (Linehan, 1993c). As discussed in Chapter 4, the constructs of emotion regulation and mindfulness are explicitly taught in DBT with psychoeducation, specific skills, and continued practice (Linehan, 1993c) and the wording in the measures uses the same language as that taught in DBT (Baer et al., 2004; Gratz & Roemer, 2004). CM does not develop the language of emotion regulation and mindfulness, and therefore one possible explanation of the findings is that the instruments using these linguistic constructs favour DBT. Only 38% of participants no longer met diagnostic criteria for BPD at post- treatment and of those identified as improved on a measure of BPD severity, only 45% no longer met the diagnostic criteria for BPD. This calls into question how feasible it is for individuals to “recover” from BPD in terms of the diagnostic criteria. In contrast, when assessing improvement according to a range of clinically relevant outcomes, on average participants improved on five out of a possible nine domains. Future research should explore on what outcomes persons with a diagnosis of BPD wish to experience change and how this compares with other stakeholders. To our knowledge, this is the first study to look at MCID in persons with BPD. Beyond analyses looking at change at a group level, we need to understand more 241 about what happens at an individual level. We have identified which outcomes were associated with greater proportions improving within 14 months. The greatest weakness of this subset of analyses is that we were only able to include participants for whom we had post-treatment data. Hence, the conclusions cannot be drawn regarding the whole sample. Further, a larger proportion of participants from the CM condition completed their post-treatment assessment than was the case for DBT participants and, as such, it makes it difficult to compare between groups. While the proportion of participants who had ceased suicide attempts was very promising, these data were based on the seven months since the previous assessment. We would need a longer timeframe to be able to claim more definitively that the behaviour had ceased. This information should be collected in follow-up studies. There are multiple ways to conceptualise and calculate MCID, including clinical consensus, distribution, and anchor-based methods. Only anchor based is truly patient-centred as it anchors the change in outcome to the patient’s subjective sense of improvement (Button et al., 2015). At this point, we do not know what constitutes a benefit that would be considered worthwhile to the patient. Norman, Sloan, and Wyrwich (2003) conducted a systematic review of studies that have computed the MCID for health related quality of life instruments. They found that in most circumstances, the threshold of discrimination was approximately half a standard deviation. Ideally, studies such as Button et al (2015) conducted to calculate MCID for depression, should occur for a BPD sample. The current study did not have sufficient power to replicate this with a BPD sample. Further research is needed to identify how 242 to encapsulate what constitutes a benefit that would be considered worthwhile in BPD. 243 Chapter 7 Overall Summary and Future Directions The preceding chapters have outlined the rationale for an RCT of Dialectical Behaviour Therapy (DBT) and Conversational Model (CM) for Borderline Personality Disorder (BPD) and presented the methodology of that RCT, an adherence tool for use in the RCT, and the outcomes. This final chapter will summarise the results, identify strengths and weaknesses of the research, and suggest future directions. 244 In the past 30 years, an increasing number of therapies designed specifically for BPD have been developed and evaluated in tightly controlled research settings (Cristea et al., 2017). However, there are few studies with strong methodology examining how well these therapies perform in real world settings or how well they compare to active treatments (as opposed to treatment as usual or waiting list control conditions). The current study is the first effectiveness study using an RCT design of DBT and CM for the treatment of suicidal and non-suicidal self-injurious behaviour and depression severity among persons with BPD. This study aimed to extend our knowledge about how well the potential benefits of DBT translate to public mental health treatment settings, with therapists in this trial being routinely employed by the health service. Whilst there have been a number of studies of DBT compared with treatment as usual (TAU) or a waiting list control, fewer studies have compared DBT with an active treatment. In the present study CM was selected as the active treatment control condition as it is one of the few treatments specifically designed for the treatment of BPD that is taught and used in Australia (Korner & McLean, 2017). Although CM has been the focus of a small number of studies, none used a randomised design. It was decided not to conduct a randomised study comparing CM against TAU as the existing data appeared sufficiently promising to believe that this might be an effective treatment and so we moved to a comparison of CM against an established active treatment such as DBT. This discussion brings together the findings of the six previous chapters, which comprise the thesis, and places them in the context of the thesis aims. Strengths and limitations of the research will also be discussed, and recommendations for future research will be presented. 245 7.1 Key Findings of the Thesis This thesis included six chapters outlining the rationale for an RCT of DBT and CM for the treatment of BPD, the methodology of the RCT, the outcomes from the trial, and the development of an adherence tool for CM. The aims of the research were to: 1. Develop a psychometrically valid measure to test adherence to CM. 2. Evaluate the effectiveness of CM against DBT, in a routine clinical setting. 3. Compare CM and DBT in relation to the working alliance. 4. Explore improvement and deterioration at the individual patient level. Chapter 1 outlined the published outcome research of the psychotherapeutic treatment of BPD. This opening chapter established the empirical basis of DBT and CM in the treatment of BPD in adults. There are now a number of evidence-based therapies for BPD, of which DBT has the most empirical evidence. RCTs of BPD have mostly been conducted in tightly controlled research settings (Tables 1.1-1.3). While several studies have evaluated DBT in routine clinical settings, as shown in Tables 1.4- 1.6, most of these include various sources of bias, such as non-randomisation, non- blind assessments, lack of fidelity assessment, and small sample sizes (see Table 1.7). It was expected when planning the current study that both treatments would lead to significant improvement across a range of domains and specifically, that DBT would be more effective than CM in reducing episodes of suicidal and NSSI and CM would be more effective than DBT in reducing depression severity. Chapter 2 described the research protocol/methodology of the RCT conducted at the Centre for Psychotherapy, providing information about the study setting as part of public mental health in Newcastle, Australia. The chapter also provided information 246 about inclusion and exclusion criteria, the content of the interventions, the therapists’ backgrounds and the training provided, as well as the overall procedure. Finally, detailed information was provided about the sample size calculation and the plan for the data analysis to be used in Chapter 4. In the absence of an available adherence tool to measure adherence to CM, Chapter 3 outlined the development of the Newcastle Adherence Scale for CM (NASCOM). The tool was found to have good psychometric properties and to adequately discriminate between CM sessions and DBT sessions. The NASCOM was used to assess adherence in the RCT. Results of the RCT were presented in Chapter 4. Both treatments showed significant improvement over time across the 14 months of therapy in suicidal and NSSI and depression scores. However, contrary to expectation, there was no significant difference between the treatment models in the reduction of suicidal and NSSI and DBT was associated with significantly greater reductions in depression scores compared to CM. The pattern of results was similar with the secondary outcomes such that scores on BPD severity, dissociation, interpersonal problems, sense of self, mindfulness capacity and difficulties in emotion regulation all significantly improved with both treatment conditions. However, DBT was associated with significant improvement in mindfulness capacity and emotion regulation compared to CM. Overall, our results are consistent regarding clinical improvements with those of Meares’ team who conducted the outcome studies for CM (Korner et al., 2006; Stevenson & Meares, 1992; Stevenson et al., 2005). In line with Comtois et al. (2007), if we benchmark our scores against key efficacy studies (see Table 1.3), our findings are 247 similar for reducing suicidal and NSSI as well as other areas of psychopathology. However, our dropout rates (44% for DBT and 41% for CM) were higher than most of the efficacy studies outlined in Chapter 1. Interestingly, our results are similar to an earlier study conducted by Linehan et al. (2002) which compared DBT with a treatment called “Comprehensive Validation Therapy” (CVT) in a small sample of 23 opioid dependent women who also met DSM criteria for BPD. This was one of the few studies where DBT performed similarly to the control condition and did not significantly reduce suicide and NSSI beyond the control condition. CVT was a treatment designed for the control arm of the study, which involved the acceptance components of DBT without the change components. Both DBT and CVT were associated with significant reductions in opiate use and overall levels of psychopathology, with CVT having no dropout from treatment compared with participants in the DBT condition where the dropout rate was 36%. The findings from that study were inconsistent with the philosophy of DBT (Linehan, 1993a) that posits that change-based strategies are an essential part of treatment to facilitate improvement. The current study also challenges that DBT proposition given that both treatments were associated with significant reductions in suicidal and NSSI, despite CM not including any active change-based strategies. The finding that depression severity improved significantly more in the DBT arm than CM was unexpected. It is possible that the change strategies from DBT were helpful in improvements in depression scores. The Linehan et al. (2002) study described above did not measure depression severity. Other studies evaluating DBT have had mixed results regarding change in depression severity (Stoffers et al., 2012). Future replications are needed of studies that compare treatments with acceptance 248 and change-based strategies in the treatment of depression severity in BPD with those that use acceptance-based strategies alone. Chapter 5 reported on the findings of the RCT, in relation to the working alliance. We compared the therapeutic alliance in DBT and CM, using therapist-rated and client-rated estimates of alliance as well as sub-scale scores of the alliance in the task, goal, and bond. Consistent with previous research that found DBT to be associated with superior alliance ratings compared to a control condition (Bedics et al., 2015), we expected greater alliance ratings in DBT than CM. We also expected greater task and goal ratings in DBT than CM, reflecting the emphasis of these components in DBT. Overall, we predicted a significant positive relationship between alliance ratings and therapeutic outcome in both therapies. We found that there was a treatment effect overall for client-rated goals, which was significantly greater in DBT than CM. Scores on therapist-rated goals were also significantly greater in DBT than CM in a time by treatment interaction effect. In terms of the relationship between alliance ratings and therapeutic outcome, the only outcome variables that showed a significant relationship with client-rated alliance were depression scores and interpersonal problems, such that higher ratings of alliance were associated with greater improvement on depression scores and interpersonal problems. Chapter 6 investigated changes at the individual level, beyond the aggregated results by treatment group reported in Chapters 4 and 5. The majority of participants improved in terms of their suicidal and non-suicidal self-injury, the severity of BPD symptoms, depression scores, and dissociation scores. Despite this, only 38% of the 249 sample no longer met criteria for BPD and the majority of participants were unchanged on interpersonal problems, sense of self, and mindfulness capacity. 7.2 Integration of Key Findings of the Thesis Key findings in this thesis included the following: 1. BPD can be successfully treated in routine clinical settings for key clinical outcomes. 2. CM can be reliably assessed for adherence to the treatment model. 3. Both DBT and CM were associated with significant improvements in suicide attempts and NSSI. CM on most outcomes, produced comparable results to DBT with a similar level of fidelity and treatment retention. DBT was associated with significantly better improvements than CM for depression severity. Hence, DBT should be considered in preference to CM for individuals with co-morbid BPD and depression. 4. Higher client-rated therapeutic alliance was associated with improved outcomes on depression severity and interpersonal problems. 5. Most participants continued to meet diagnostic criteria for BPD after 14 months of treatment. 6. The majority of participants improved on suicidal and non-suicidal self-injury, depression severity, and BPD severity. Nevertheless, the majority remained unchanged after 14 months of therapy on domains of interpersonal problems, sense of self, and mindfulness capacity. 250 7.3 Strengths and Limitations 7.3.1 Strengths This thesis contributes substantially to the knowledge base regarding treatment for BPD in routine clinical settings. Stoffers et al. (2012), in the recent Cochrane review of psychological therapies for BPD, suggest “there is an urgent need for independent research endeavours” (p. 77) and the trial outlined in this thesis goes some way to providing such evidence, given that none of the investigators were involved in the development of DBT or CM, nor have they worked alongside any of the treatment developers. It has several strengths, including the careful and systematic development of a psychometrically valid adherence tool for CM, an RCT design that was adequately powered to detect a clinically significant difference and examination of clinical improvement and deterioration at an individual level. The results of this thesis have demonstrated that it is possible to treat persons with BPD using either CM or DBT in a routine clinical setting. The service where the study was conducted had already been the setting of an RCT comparing six months of DBT with a wait list control condition (Carter et al., 2010). As such, the feasibility of running a clinical trial using a RCT design at this service had already been established. The sample size for the current trial was sufficiently large to provide adequate power to examine differential changes in the primary outcomes. Outcomes were assessed using well-validated measures commonly used in other BPD trials. A number of other elements in the trial were designed to reduce potential bias and hence strengthen the validity (Cristea et al., 2017). The stratified randomisation based on gender and antidepressant medication controlled for the possible impact of 251 these two factors on the outcomes. Blindness of raters for both the assessment of outcomes and for adherence ratings further reduced potential bias of outcomes. Intention to treat analyses were conducted and missing data was controlled for by the use of Generalized Linear Models. Adherence coding was conducted for both treatments to maximise confidence that therapists were delivering the treatment intended and confirmed that therapy sessions were largely adherent to the treatment model. The adherence coding commenced after the initiation of the trial, and therapists did not receive the results of this coding at any stage during the trial. It was decided not to provide therapists with their adherence results as such feedback would generally not occur in a real world setting. Feedback provided mid-way through the trial would have introduced a confounding variable by leading to an improvement in adherence to the model and potentially to outcomes. This may undermine our capacity to obtain an accurate picture of the effectiveness of the interventions in a routine clinical practice setting. Stoffers et al. (2012) criticised the majority of trials which have investigated the efficacy of psychological therapies in BPD, on the basis that control participants were not given comparable amounts of professional attention compared to participants in the treatment groups. In the current study, participants in the CM condition attended twice-weekly individual therapy while those in the DBT condition attended once weekly individual therapy and once weekly skills training. As such, the amount of attention received from professionals was comparable, even though those in the CM group received two hours of face-to-face individual therapy and those in the DBT group received one hour of face-to-face individual therapy and 2.5 hours of skills training (with two therapists and usually eight people in a group). Ideally, a third arm, 252 such as a minimal treatment control group, would have been included. However, it was deemed unethical for persons at known risk of suicide to be denied an evidence- based psychotherapeutic treatment when one was available. 7.3.2 Limitations The main weakness of the current RCT was participant retention in the study. Unlike most efficacy trials, there was little incentive for participants to attend assessments as they received treatment regardless of their participation in the study and attendance at assessments. Participants were compensated a small amount (AUD $20) for transport costs to attend the assessments, sent reminder text messages the day before, sent personalised thank you cards after they had attended the assessments, and welcomed into a relaxing environment at the assessments by providing water, tea or coffee and sweets, i.e., confectionery. Most participants (71%) completed the post-assessments. This was an acceptable amount of attrition for ITT analyses to take into account the missing data, however, it does prevent us from being able to comment definitively on the functioning of the remaining participants. There was no funding for this research. A small grant was obtained while the study was underway (AUD $10,000) which was used to pay for the DBT adherence coding. Following completion of recruitment additional small grants were obtained to pay for statistical consultation, research assistant hours and attendance at international research conferences. The author of this thesis ran the research project while also working full-time on a clinical basis. A research assistant was employed for three days per week (and paid from money earned by running training in DBT or CM by clinicians of the team). This is a fraction of the research assistant time that most large trials would have and meant that we were somewhat limited in our ability to “go the 253 extra mile” to improve participant retention in the study, for example, going out to meet the patients who did not want to travel to the clinic for their assessments. Further, for reasons unclear to us, the participant retention in the study was uneven between the two treatment conditions (CM = 75% retention; DBT = 65% retention). One aspect that was different between the two treatments was the group element, with only participants in DBT interacting with other participants in the group skills training as part of the treatment. One speculation regarding the discrepancy in attendance of the research assessments between participants in the two conditions is that DBT participants communicated something about their experience of the assessment with each other during group skills training that led to discouragement of attendance at the research assessments. Another possibility is that therapists offering CM may have been more supportive of the research overall (and hence, emotionally supported their patients to attend the research assessments). However, these are mere speculations, and qualitative research in future may help us to understand the factors that have a bearing on attendance at assessments. Although the Centre where the study was undertaken sits within public sector mental health services, it is a specialist service for BPD and Eating Disorders. As such, while the therapists were not specifically hired for the study, they are representative of therapists found in real world settings, however, they have expertise in BPD beyond that which may be found in a general community service. Hence, the generalisability of the findings to general treatment settings may be limited. 254 7.4 Future Directions This thesis has addressed some gaps in the literature and highlighted the need for a number of future research directions. This study extends the efficacy trial results on the treatment of BPD in routine clinical settings and has empirically shown that it is possible to treat BPD using DBT or CM in an effective way outside of tightly controlled research settings. The current research has also highlighted a number of areas that require consideration for further research and these will be discussed below. The current study is the first evaluation of CM in a randomised design. The outcomes from the current study are superior to those achieved with wait list or control conditions (see Table 1.3 and Table 1.6). Regardless, replication of CM in another randomised study is necessary to confirm the results of the current study. Future research should incorporate longer follow-up periods to ensure that the gains achieved at post-therapy are maintained and indeed, whether they continue to improve. Two studies of MBT described in Chapter 1 (Bateman et al, 1999; Bateman et al, 2009) found that the significant changes between MBT and the control condition did not occur until the final six months of the 18 month trial period. It is possible that the effects of CM would be more evident beyond the 14 month trial period. Longer follow-up periods are necessary to identify this. Many of the existing BPD outcome studies have short follow-up periods relative to the length of treatment and disturbance of the disorder, with the longest reported follow-up period being 2 years. This thesis included a discussion of the importance of examining clinically significant change for persons with a diagnosis of BPD and suggested the use of Minimally Clinically Important Difference (MCID) as one option to capture the patient’s subjective sense of improvement. Further studies should examine what would 255 constitute a benefit that patients would consider worthwhile. Using this information, data from larger samples should then be collected to calculate MCID for BPD. BPD psychological intervention trials have higher treatment dropout rates than trials of psychological interventions for individuals with Axis I disorders (Kelly et al., 1992). The treatment dropout rates in our study were higher than those reported in systematic reviews of efficacy studies of BPD (Cristea et al., 2017), but fairly standard within the range of those reported in effectiveness studies of BPD. There are multiple reasons why people may drop-out of treatment. Previous studies of factors associated with treatment drop-out from DBT include younger age, higher levels of baseline distress and higher level of baseline non-acceptance of emotions (Landes, Chalker & Comtois, 2016) as well as higher levels of anger, greater Axis I co-morbidity and higher number of lifetime suicide attempts (Wnuk et al, 2013). We know from previous studies that specialised therapies for BPD are associated with greater improvements than those who receive TAU. Hence, we could assume that those who drop out and hence, receive a lesser dose of the therapy may do less well than those who stay in treatment. Further research is needed to explore ways to reduce the dropout rate and to maintain people in treatment. Different psychotherapies have different theoretical explanations of the aetiology of the disorder that informs the development and delivery of treatment. Interestingly, despite differing protocols and perspectives on aetiology of the disorder of BPD, DBT and CM produced similar outcomes on a range of variables with the majority of outcomes showing no significant difference between the two treatments. Further analyses will examine mediation of treatment outcome to identify the specific treatment processes for each model that were associated with change. 256 In addition, there is little research investigating which components of DBT actively contribute to its therapeutic effect. The protocol for Standard DBT includes individual therapy, skills training groups, after hours phone coaching, and consultation team meetings for therapists and as such is very resource intensive. To date, there has only been one “dismantling study” (Linehan et al., 2015) that has examined different components of the model. That study included three arms: Standard DBT that included all of the components (skills training group, individual therapy, phone coaching and therapist consultation); skills training group + case management; or DBT individual therapy + activities group. The results were mixed in that all arms were associated with improvements in suicide attempts, use of crisis services, and reasons for living. However, the Standard DBT arm of the trial had significantly lower dropout rates and better outcomes one year after completing treatment. Replication studies investigating active components are required. Phone coaching is one component of DBT that has had very little investigation regarding the contribution it makes to outcomes. Chalker et al. (2015) found more frequent phone contacts in DBT (regardless of during or outside of the work day) were associated with a decrease in dropout and psychological symptoms, and an increase in client and therapist satisfaction in DBT. However, there is no research currently in the DBT literature on the impact of phone coaching hours. Outside of DBT, Nadort et al. (2009) in a study of Schema Therapy which randomised patients to receive phone support from their individual therapist after hours or not, found that access out-of- hours made no difference to outcomes. In the current study, participants in the DBT condition had access to phone coaching out of hours, whereas those in the CM condition did not. It is not known what impact this had on the outcomes. It is possible 257 it may account for the significant difference in the outcome of depression severity between the two treatments, given that extra support was available. Further research is needed to determine the additive impact of having access to phone support after hours. Both therapies were very labour intensive and may not be feasible for many services to implement (Landes et al., 2017). There are a number of studies conducted in routine clinical settings reporting benefit of six months of DBT (Ben-Porath et al., 2004; Blennerhassett et al., 2009; Brassington & Krawitz, 2006; Carter et al., 2010; Pasieczny & Connor, 2011; Prendergast & McCausland, 2007; Williams et al., 2010). However, most of these studies were uncontrolled and had small sample sizes. Future research should investigate whether there is additional benefit from the longer length of therapy, (i.e., six vs. 14 months as in the current study) and if so, the extent of any additional benefit. The current research has shown that in a routine clinical setting, the majority of individuals will obtain improvement in terms of life-interfering behaviours and depression scores. However, some individuals stayed the same and a small number deteriorated. Any treatment that is effective is unlikely to be effective for everybody (McMain et al., 2015). When there is more than one evidence-based treatment available, clinicians should have the ability to select the treatment that is likely to be the best match for each patient. There is growing interest in this approach in medical research, and it has been referred to as personalised or precision medicine, where the aim is to develop individually tailored treatments for patients (Silberschatz, 2017). To date, in the field of BPD, there is little empirical evidence to guide clinicians in making that decision, and this is a critical area where future research is needed. Moderation 258 analyses seek to determine for whom or under what conditions treatment choice differentially affects outcome (Kraemer, 2016). For the current study, moderation analyses will be conducted to determine if we are able to identify moderators that will assist us to better target treatment to the specific symptom profile of each person. In the current research, DBT had better outcomes in terms of depression scores. It may be that depression at baseline is a moderator of treatment outcome. Successfully identifying the moderators of treatment outcome, will allow us to move closer to being able to match the therapy to the patient. 7.5 Conclusion While the field has made much progress in identifying evidence-based treatments for BPD in the past 30 years (Cristea et al., 2017) and there is now broad consensus that the disorder is common and can be treated (Leichsenring et al., 2011), there remains more work to be done for all patients to obtain a good outcome across treatment settings (Paris, 2012). The research described in this thesis has contributed to the current knowledge in the field of treatment outcome for BPD. It has also raised further areas for us to explore in order to improve our understanding and ultimately, improve the outcomes for individuals seeking help for this disabling disorder. 259 References American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington: DC: Psychiatric Press. Arntz, A., van de Hoorn, M., Cornelius, J., Verheul, R., van den Bosch, W., & de Bie, A. (2003). Reliability and validity of the Borderline Personality Disorder severity index. Journal of Personality Disorders, 17, 45–59. Atkins, M. S., Frazier, S. L., & Cappella, E. (2006). Hybrid research models: Natural opportunities for examining mental health in context. Clinical Psychology: Science and Practice, 13(1), 105–108. Aviram, R. B., Brodsky, B. S., & Stanley, B. (2006). Borderline Personality Disorder, stigma, and treatment implications. Harvard Review of Psychiatry, 14(5), 249– 256. doi:10.1080/10673220600975121 Baer, B. A., Smith, G. T., & Allen, K. B. (2004). Assessment of mindfulness by self- report: The Kentucky Inventory of Mindfulness Skills. Assessment, 11, 191–206. Baker, A.L., Hiles, S.A., Thornton, L.K., Hides, L., Lubman, D.I. (2012). A systematic review of psychological interventions for excessive alcohol consumption among people with psychotic disorders. Acta Psychiatrica Scandanavia, 126, 243-255. Baker, A. L., Kavanagh, D. J., Kay-Lambkin, F. J., Hunt, S. A., Lewin, T. J., Carr, V. J., & McElduff, P. (2014). Randomized controlled trial of MICBT for co-existing alcohol misuse and depression: Outcomes to 36 months. Journal of Substance Abuse Treatment, 46, 281–290. 260 Barber, J. P., Connolly, M. B., Crits-Christoph, P., Gladis, L., & Siqueland, L. (2009). Alliance predicts patients’ outcome beyond in-treatment change in symptoms. Personality Disorders: Theory, Research, and Treatment, S(1), 80-89. Barber, J. P., Liese, B. S., & Abrams, M. J. (2003). Development of the cognitive therapy adherence and competence scale. Psychotherapy Research, 13, 205–221. Barnicot, K., Katsakou, C., Bhatti, N., Savill, M., Ferns, N., & Priebe, S. (2012). Factors predicting the outcome of psychotherapy for Borderline Personality Disorder: A systematic review. Clinical Psychology Review, 32(6), 400–412. doi:10.1016/j.cpr.2012.04.004 Basten, C. J. (2008). Development and validation of a measure of subjective sense of self: The Sense of Self Inventory (Unpublished doctoral thesis). University of Sydney, Australia. Bateman, A., & Fonagy, P. (1999). Effectiveness of partial hospitalization in the treatment of Borderline Personality Disorder: A randomized controlled trial. American Journal of Psychiatry, 156(10), 1563–1569. doi:http://dx.doi.org/10.1176/ajp.156.10.1563 Bateman, A. W., & Fonagy, P. (2000). Effectiveness of psychotherapeutic treatment of Personality Disorder. British Journal of Psychiatry, 177, 138–143. Bateman, A., & Fonagy, P. (2009). Randomized controlled trial of outpatient Mentalization-Based Treatment versus structured Clinical Management for Borderline Personality Disorder. American Journal of Psychiatry, 166(12), 1355- –1364. Bateman, A. W., Gunderson, J., & Mulder, R. (2015). Treatment of personality disorder. The Lancet, 385(9969), 735-743. 261 Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Manual for Beck Depression Inventory II. San Antonio, TX: Psychological Corporation. Beck, A. T., Steer, R. A., & Carbin, M.G. (1988). Psychometric properties of the Beck Depression Inventory: Twenty-five years of evaluation. Clinical Psychology Review, 8, 77-100. Bedics, J. D., Atkins, D. C., Harned, M. S., & Linehan, M. M. (2015). The therapeutic alliance as a predictor of outcome in dialectical behavior therapy versus nonbehavioral psychotherapy by experts for Borderline Personality Disorder. Psychotherapy (Chic), 52(1), 67–77. doi:10.1037/a0038457 Ben-Porath, D. D., Peterson, G. A., & Smee, J. (2004). Treatment of individuals with Borderline Personality Disorder using Dialectical Behavior Therapy in a community mental health setting: Clinical application and a preliminary investigation. Cognitive and Behavioral Practice, 11(4), 424–434. doi:http://dx.doi.org/10.1016/S1077-7229%2804%2980059-2 Bernstein, D. P., & Putman, F. W. (1986). Development, reliability, validity of a dissociation scale. Journal of Nervous and Mental Disease, 174(12), 727–735. Black, D. W., Blum, N., Eichinger, L., McCormick, B., Allen, J., & Sieleni, B. (2008). STEPPS: Systems Training for Emotional Predictability and Problem Solving in women offenders with Borderline Personality Disorder in prison: A pilot study. CNS Spectrums, 13(10), 881–886. Black, D. W., Blum, N., Pfohl, B., & Hale, N. (2004). Suicidal behaviour in Borderline Personality Disorder: Prevalence, risk factors, prediction and prevention. Journal of Personality Disorders, 18, 226–239. 262 Blennerhassett, R., Bamford, l., Whelan, A., Jamieson, S., & O’Raghaillaigh, J. W. (2009). Dialectical Behaviour Therapy in an Irish community mental health setting. Irish Journal of Psychological Medicine, 26(2), 59–63. Blum, N., St. John, D., Pfohl, B., Stuart, S. P., McCormick, B., Allen, J., . . . Black, D. W. (2008). Systems Training for Emotional Predictability and Problem Solving (STEPPS) for outpatients with Borderline Personality Disorder: A randomized controlled trial and 1-year follow-up. American Journal of Psychiatry, 165(4), 468–478. Bohus, M., Haaf, B., Simms, T., Limberger, M. F., Schmahl, C., Unkel, C., . . . Linehan, M. M. (2004). Effectiveness of inpatient dialectical behavioral therapy for Borderline Personality Disorder: A controlled trial. Behaviour Research and Therapy, 42, 487–499. doi:10.1016/S0005-7967(03)00174-8 Bongar, B., Peterson, L. G., Golann, S., & Hardiman, J.J. (1990). Self-mutilation and the chronically “suicidal” emergency room patient. Annals of Clinical Psychiatry, 2, 217–222. Bordin, E. (1975, August). The working alliance: Basis for general theory of psychotherapy. Paper presented at the annual meeting of the American Psychological Association, Washington, DC. Bordin, E. S. (1979). The generalizability of the psychoanalytic concept of the working alliance. Psychotherapy: Theory, Research & Practice, 16(3), 252-260. Borkovec, T. D., & Castonguay, L. G. (1998). What is the scientific meaning of empirically supported therapy? Journal of Consulting and Clinical Psychology, 66(1), 136–142. 263 Bos, E. H., Bas van Wel, E., Appelo, M. T., & Verbraak, M. J. P. M. (2010). A randomized controlled trial of a Dutch version of Systems Training for Emotional Predictability and Problem Solving for Borderline Personality Disorder. Journal of Nervous and Mental Disease, 198(4), 299–304. Brassington, J., & Krawitz, R. (2006). Australasian Dialectical Behaviour Therapy pilot outcome study: Effectiveness, utility and feasibility. Australasian Psychiatry, 14(3), 313–319. doi:10.1111/j.1440-1665.2006.02285.x Brown, H., & Prescott, R. (Eds.). (2006). Applied mixed models in medicine. England: Wiley & Sons. Budge, S. L., Moore, J. T., Del Re, A. C., Wampold, B. E., Baardseth, T. P., & Nienhuis, J. B. (2013). The effectiveness of evidence-based treatments for personality disorders when comparing treatment-as-usual and bona fide treatments. Clinical Psychology Review, 33, 1057–1066. Burnham, K. P., & Anderson, D. R. (2004). Multimodel inference understanding AIC and BIC in model selection. Sociological Methods and Research, 33(2), 261–304. doi:10.1177/0049124104268644 Butler, S. F., Henry, W. P., & Strupp, H. H. (1995). Measuring adherence in time-limited dynamic psychotherapy. Psychotherapy (Chic), 32, 629–638. Button, K., Kounali, D., Thomas, L., Wiles, N., Peters, T., Welton, N., Ades, A. & Lewis, G. (2015). Minimal clinically important difference on the Beck Depression Inventory-II according to the patient’s perspective. Psychological Medicine, 45(15), 3269–3279. doi:http://dx.doi.org/10.1017/S0033291715001270 264 Carroll, K. M., & Rounsaville, B. J. (2003). Bridging the gap: A hybrid model to link efficacy and effectiveness research in substance abuse treatment. Psychiatric Services, 54(3), 333–339. Carter, G. L., Willcox, C. H., Lewin, T. J., Conrad, A. M., & Bendit, N. (2010). Hunter DBT project: Randomized controlled trial of Dialectical Behaviour Therapy in women with Borderline Personality Disorder. Australian and New Zealand Journal of Psychiatry, 44(2), 162–173. doi:http://dx.doi.org/10.3109/00048670903393621 Chalker, S. A., Carmel, A., Atkins, D. C., Landes, S. J., Kerbrat, A. H., & Comtois, K. A. (2015). Examining challenging behaviors of clients with Borderline Personality Disorder. Behaviour Research and Therapy, 75, 11–19. Chambless, D. L., & Hollon, S. D. (1998). Defining empirically supported therapies. Journal of Consulting and Clinical Psychology, 66(1), 7–18. Chambless, D. L., & Ollendick, T. H. (2001). Empirically supported psychological interventions: Controversies and evidence. Annual Review of Psychology, 52, 685–716. Chan, L. S. (2013). Minimal Clinically Important Difference (MCID)—Adding meaning to statistical inference. American Journal of Public Health, 103, 11(24–25). Chanen, A. M. (2015). Borderline personality disorder in young people: are we there yet?. Journal of clinical psychology, 71(8), 778-791. Choi-Kain, L. W., Albert, E. B., & Gunderson, J. G. (2016). Evidence-based treatments for Borderline Personality Disorder: Implementation, integration, and stepped care. Harvard Review of Psychiatry, 24(5), 342–356. doi:http://dx.doi.org/10.1097/HRP.0000000000000113 265 Clarkin, J. F., Levy, K. N., Lenzenweger, M. F., & Kernberg, O. F. (2007). Evaluating three treatments for Borderline Personality Disorder: A multiwave study. American Journal of Psychiatry, 164(6), 922–928. Comtois, K. A., Elwood, L., Holdcraft, L. C., Smith, W. R., & Simpson, T. L. (2007). Effectiveness of dialectical behavior therapy in a community mental health center. Cognitive and Behavioral Practice, 14(4), 406–414. doi:http://dx.doi.org/10.1016/j.cbpra.2006.04.023 Cottraux, J., Druon Note, I., Boutitie, F., Milliery, M., Genouihlac, V., Yao, S. N., . . . Gueyffier, F. (2009). Cognitive Therapy versus Rogerian Supportive Therapy in Borderline Personality Disorder Two-Year Follow-Up of a Controlled Pilot Study. Psychotherapy Psychosom, 78, 307-316. Cristea, I. A., Gentili, C., Cotet, C. D., Palomba, D., Barbui, C., & Cuijpers, P. (2017). Efficacy of psychotherapies for Borderline Personality Disorder: A systematic review and meta-analysis. JAMA Psychiatry, 74(4), 319–328. Crits-Christoph, P., Gibbons, M. B. C., Hamilton, J., Ring-Kurtz, S., & Gallop, R. (2011). The dependability of alliance assessments: The alliance–outcome correlation is larger than you might think. Journal of Consulting and Clinical Psychology, 79(3), 267–278. Davidson, K., Norrie, J., Tyrer, P., Gumley, A., Tata, P., Murray, H., & Palmer, S. (2006a). The effectiveness of cognitive behavior therapy for borderline personality disorder: results from the borderline personality disorder study of cognitive therapy (BOSCOT) trial. Journal of Personality Disorders, 20(5), 450-465. Davidson, K., Tyrer, P., Gumley, A., Tata, P., Norrie, J., Palmer, S., . . . Macauley, F. (2006b). Cognitive Behavior Therapy for Borderline Personality Disorder: 266 Rationale for Trial, Method, and Description of Sample. Journal of Personality Disorders, 20(5), 431-449. Davidson, K. M., Tyrer, P., Norrie, J., Palmer, S. J., & Tyrer, H. (2010). Cognitive therapy v. usual treatment for borderline personality disorder: prospective 6-year follow-up. The British Journal of Psychiatry, 197(6), 456-462. Davidson, L., & Roe, D. (2007). Recovery from versus recovery in serious mental illness: One strategy for lessening confusion plaguing recovery. Journal of Mental Health, 16(4), 459-470. Doering, S., Horz, S., Rentrop, M., Fischer-Kern, M., Schuster, P., Benecke, C., . . . Buchheim, P. (2010). Transference-focused psychotherapy v. treatment by community psychotherapists for Borderline Personality Disorder. British Journal of Psychiatry, 196, 389–395. Doran, J. M., Safran, J. D., Waizmann, V., Bolger, K., & Muran, J. C. (2012). The alliance negotiation scale: Psychometric construction and preliminary reliability and validity analysis. Psychotherapy Research, 22(6), 710–719. doi:10.1080/10503307.2012.709326 Ekers, D., Webster, L., Van Straten, A., Cuijpers, P., Richards, D., & Gilbody, S. (2014). Behavioural activation for depression: An update of meta-analysis of effectiveness and sub group analysis. Plos One, 9(6), e100100. Faerstein, I., & Levenson, H. (2016). Validation of a fidelity scale for accelerated- experiential dynamic psychotherapy. Journal of Psychotherapy Integration, 26(2), 172–185. Farrell, J. M., Shaw, I. A., & Webber, M. A. (2009). A schema-focused approach to group psychotherapy for outpatients with Borderline Personality Disorder: A 267 randomized controlled trial. Journal of Behavior Therapy and Experimental Psychiatry, 40, 317–328. doi:10.1016/j.jbtep.2009.01.002 Feigenbaum, J. D., Fonagy, P., Pilling, S., Jones, A., Wildgoose, A., & Bebbington, P. E. (2012). A real-world study of the effectiveness of DBT in the UK National Health Service. British Journal of Clinical Psychology, 51(2), 121–141. doi:http://dx.doi.org/10.1111/j.2044-8260.2011.02017.x First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B. W. (1995). Structured clinical interview for Axis I DSM-IV disorders-patient edition (SCID-I/P). New York: New York State Psychiatric Institute Biometrics Research Department. Flay, B. R., Biglan, A., Boruch, R. F., González Castro, F., Gottfredson, D., Kellam, S., . . . Ji, P. (2005). Standards of evidence: Criteria for efficacy, effectiveness and dissemination. Prevention Science, 6(3), 151–175. Fluckiger, C., Del Re, A. C., Wampold, B. E., Symonds, D., & Horvath, A. O. (2012). How central is the alliance in psychotherapy? A multilevel longitudinal meta- analysis. Journal of Counseling Psychology, 59(1), 10–17. doi:10.1037/a0025749 Fonagy, P., & Target, M. (2006). The mentalization-focused approach to self pathology. Journal of Personality Disorders, 20(6), 544–576. Freire, E., Elliot, R., & Westwell, G. (2014). Person-centred and experiential psychotherapy scale: Development and reliability of an adherence/competence measure for person-centred and experiential psychotherapies. Counselling and Psychotherapy Research, 14(3), 220–226. Furukawa, T., Noma, H., Caldwell, D., Honyashiki, M., Shinohara, K., Imai, H., . . . Churchill, R. (2014). Waiting list may be a nocebo condition in psychotherapy 268 trials: A contribution from network meta‐analysis. Acta Psychiatrica Scandinavica, 130(3), 181-192. Gartlehner, G., Hansen, R. A., Nissman, D., Lohr, K. N., & Carey, T. S. (2006). A simple and valid tool distinguished efficacy from effectiveness studies. Journal of Clinical Epidemiology, 59(10), 1040–1048. Gaston, L. (1991). Reliability and criterion-related validity of the California Psychotherapy Alliance Scales—patient version. Journal of Consulting and Clinical Psychology, 3(1), 68–74. doi:10.1037/1040-3590.3.1.68 Giesen-Bloo, J., & Arntz, A. (2007). Outpatient psychotherapy for Borderline Personality Disorder: A randomized trial of schema-focused therapy vs. transference-focused psychotherapy: Reply. Archives of General Psychiatry, 64(5), 610–611. doi:http://dx.doi.org/10.1001/archpsyc.64.5.610 Giesen-Bloo, J., Van Dyck, R., Spinhoven, P., Van Tilburg, W., Dirksen, C., Van Asselt, T., . . . Arntz, A. (2006). Outpatient psychotherapy for Borderline Personality Disorder: Randomized trial of schema-focused therapy vs. transference-focused psychotherapy. Archives of General Psychiatry, 63, 649–658. Glasgow, R. E., Lichtenstein, E., & Marcus, A. C. (2003). Why don’t we see more translation of health promotion research to practice? Rethinking the efficacy- to-effectiveness transition. American Journal of Public Health, 93(8), 1261– 1267. Godfrey, E., Chalder, T., Ridsdale, L., Seed, P., & Ogden, J. (2007). Investigating the “active ingredients” of cognitive behaviour therapy and counselling for patients with chronic fatigue in primary care: Developing a new process measure to 269 assess treatment fidelity and predict outcome. British Journal of Clinical Psychology, 46(3), 253–272. Goldman, G. A., & Gregory, R. J. (2009). Preliminary relationships between adherence and outcome in dynamic deconstructive psychotherapy. Psychotherapy: Theory, Research, Practice, Training, 46(4), 480–485. doi:http://dx.doi.org/10.1037/a0017947 Grant, B. F., Chou, S. P., Goldstein, R. B., Huang, B., Stinson, F. S., Saha, T. D., . . . Pickering, R. P. (2008). Prevalence, correlates, disability, and comorbidity of DSM-IV Borderline Personality Disorder: Results from the Wave 2 National Epidemiologic Survey on Alcohol and Related Conditions. Journal of Clinical Psychiatry, 69(4), 533-545. Gratz, K. L., & Gunderson, J. G. (2006). Preliminary data on an acceptance-based emotion regulation group intervention for deliberate self-harm among women with Borderline Personality Disorder. Behavior Therapy, 37(1), 25–35. Gratz, K., & Roemer, L. (2004). Multidimensional assessment of emotion regulation and dysregulation: Development, factor structure and initial validation of the difficulties in emotion. Journal of Psychopathology and Behavioral Assessment, 26, 41–54. Gratz, K., Tull, M., & Levy, R. (2014). Randomized controlled trial and uncontrolled 9- month follow-up of an adjunctive emotion regulation group therapy for deliberate self-harm among women with Borderline Personality Disorder. Psychological Medicine, 44(10), 2099–2112. Gregory, R. J., Chlebowski, S., Kang, D., Remen, A. L., Soderberg, M. G., & Stepkovitch, J. (2008). A controlled trial of psychodynamic psychotherapy for co-occuring 270 Borderline Personality Disorder and Alcohol Use Disorder. Psychotherapy: Theory, Research, Practice, Training, 45(1), 28–41. doi:10.1037/0033- 3204.45.1.28 Gregory, R. J., & Ramen, A. L. (2008). A manual-based psychodynamic therapy for treatment-resistant Borderline Personality Disorder. Psychotherapy: Theory, Research, Practice, Training, 45(1), 15–27. doi:10.1037/0033-3204.45.1.15 Gunderson, J. G. (1996). Borderline patient’s intolerance of aloneness: Insecure attachments and therapist availability. American Journal of Psychiatry, 153(6), 752–758. Gunderson, J. G. (2015). Reducing suicide risk in Borderline Personality Disorder. Journal of the American Medical Association, 314(2), 181–182. Gunderson, J. G., & Links, P. S. (2008). Borderline Personality Disorder: A Clinical Guide (2nd ed.). Washington, DC: American Psychiatric Publishing. Gunderson, J. G., & Links, P. S. (2014). Handbook of good psychiatric management (GPM) for Borderline Patients. Washington, DC: American Psychiatric Press. Gunderson, J. G., Stout, R. L., McGlashan, T. H., Shea, M., Morey, L. C., Grilo, C. M., . . . Skodol, A. E. (2011). Ten-year course of Borderline Personality Disorder: Psychopathology and function from the collaborative longitudinal personality disorders study. Archives of General Psychiatry, 68(8), 827–837. doi:http://dx.doi.org/10.1001/archgenpsychiatry.2011.37 Guthrie, E. (1999). Psychodynamic interpersonal therapy. Advances in Psychiatric Treatment, 5(2), 135–145. doi:10.1192/apt.5.2.135 Haliburn, J., Stevenson, J., & Gerull, F. (2009). A university psychotherapy training program in a psychiatric hospital: 25 years of Conversational Model in the 271 treatment of patients with Borderline Personality Disorder. Australasian Psychiatry, 17(1), 25–28. Harned, M. S., Chapman, A. L., Dexter-Mazza, E. T., Murray, A., Comtois, K. A., & Linehan, M. M. (2008). Treating co-occurring Axis I disorders in recurrently suicidal women with Borderline Personality Disorder: A 2-year randomized trial of dialectical behavior therapy versus community treatment by experts. Journal of Consulting and Clinical Psychology, 76(6), 1068–1075. doi:10.1037/a0014044 Higgins, J., Green, S. (2011). Cochrane handbook for systematic reviews of interventions. Version 5.1.0. Available from: http: //handbook.cochrane.org/. Hill, C. E., O’Grady, K. E., & Elkin, I. (1992). Applying the collaborative study psychotherapy rating scale to rate therapist adherence in cognitive behavior therapy, interpersonal therapy and clinical management. Journal of Consulting and Clinical Psychology, 60(1), 73–79. Hirsh, J. B., Quilty, L. C., Bagby, R. M., & McMain, S. F. (2012). The relationship between agreeableness and the development of the working alliance in patients with Borderline Personality Disorder. Journal of Personality Disorders, 26(4), 616– 627. doi:10.1521pedi2012264616 Hjalmarsson, E., Kaver, A., Perseius, K.-I., Cederberg, K., & Ghaderi, A. (2008). Dialectical behaviour therapy for Borderline Personality Disorder among adolescents and young adults: Pilot study, extending the research findings in new settings and cultures. Clinical Psychologist, 12(1), 18–29. doi:http://dx.doi.org/10.1080/13284200802069035 Hobson, R. F. (1985). Forms of feelings: The heart of psychotherapy. London: Tavistock. 272 Hoffmann, T. C., Glasziou, P. P., Boutron, I., Milne, R., Perera, R., Moher, D., . . . Johnston, M. (2014). Better reporting of interventions: template for intervention description and replication (TIDieR) checklist and guide. BMJ, 348, g1687, 1-12. Horowitz, L. M., Rosenberg, S. E., Baer, B. A., Ureno, G., & Villasenor, V. S. (1988). Inventory of interpersonal problems: Psychometric properties and clinical applications. Journal of Consulting and Clinical Psychology, 56, 885–892. Horvath, A. O. (2000). The therapeutic relationship: From transference to alliance. Psychotherapy in Practice, 56(2), 163–173. Horvath, A. O., & Bedi, R. (2002). The alliance. In J. C. Norcross (ed.), Psychotherapy relationships that work: Therapist contributions and responsiveness to patients (pp. 37-69). Oxford: Oxford University Press. Horvath, A. O., Del Re, A. C., Fluckiger, C., & Symonds, D. (2011). Alliance in individual psychotherapy. Psychotherapy (Chic), 48(1), 9–16. doi:10.1037/a0022186 Horvath, A. O., Greenberg, L. S., & Pinsoff, W. M. (1986). The development of the working alliance inventory. In L. S. P. Greenberg & W.M. Pinsoff (Eds.), The psychotherapeutic research handbook (pp. 529–556). New York: NY: Guildford Press. Horvath, A. O., & Greenberg, L. S. (1989). Development and validation of the Working Alliance Inventory. Journal of Counseling Psychology, 36, 223–233. Horvath, A. O., & Symonds, B. D. (1991). Relation between working alliance and outcome in psychotherapy: A meta-analysis. Journal of Counselling Psychology, 38, 139–149. doi: 10.1037/0022-0167.38.2.139 273 Howard, K. I., Moras, K., Brill, P. L., Martinovich, Z., & Lutz, W. (1996). Evaluation of psychotherapy: Efficacy, effectiveness and patient progress. American Psychologist, 51(10), 1059–1064. Jacobson, N. S., Follette, W. C., & Revenstorf, D. (1984). Psychotherapy outcome research: Methods for reporting variability and evaluating clinical significance. Behavior Therapy, 15(4), 336-352. Jacobson, N. S., & Truax, P. (1991). Clinical significance: a statistical approach to defining meaningful change in psychotherapy research. Journal of consulting and clinical psychology, 59(1), 12. Jacobson, N. S., Roberts, L. J., Berns, S. B., & McGlinchey, J. B. (1999). Methods for defining and determining the clinical significance of treatment effects: Description, application, and alternatives. J Consult Clin Psychol, 67(3), 300-307. Jorgensen, C. R., Freund, C., Boye, R., Jordet, H., Andersen, D., & Kjolbye, M. (2013). Outcome of mentalization-based and supportive psychotherapy in patients with Borderline Personality Disorder: A randomized trial Acta Psychatrica Scandinavica, 127, 305–317. Kapur, N., Cooper, J., O’Connor, R. C., & Hawton, K. (2013). Non-suicidal self-injury v. attempted suicide: New diagnosis or false dichotomy? British Journal of Psychiatry, 202(5), 326–328. doi:10.1192/bjp.bp.112.116111 Karterud, S., Pedersen, G., Engen, M., Johansen, M., Johansson, P., Schluter, C., . . . Bateman, A. W. (2013). The MBT Adherence and Competence Scale (MBT-ACS): Development, structure and reliability. Psychotherapy Research, 23(6), 705– 717. 274 Katsakou, C., Marougka, S., Barnicot, K., Savill, M., White, H., Lockwood, K., & Priebe, S. (2012). Recovery in Borderline Personality Disorder (BPD): A qualitative study of service users’ perspectives. Plos One, 7(5), 1–8. Kazdin, A. E. (1999). The meanings and measurement of clinical significance. Journal of Consulting and Clinical Psychology, 67(3), 332-339. Kazdin, A. E. (2001). Almost clinically significant (p<. 10): Current measures may only approach clinical significance. Clinical Psychology: Science and Practice, 8(4), 455-462. Kelly, T., Soloff, P. H., Cornelius, J., George, A., Lis, J. A., & Ulrich, R. (1992). Can we study (treat) borderline patients? Attrition from research and open treatment. Journal of Personality Disorders, 6, 417–433. Kleindienst, N., Limberger, M. F., Ebner-Priemer, U. W., Keibel-Mauchnik, J., Dyer, A., Berger, M., . . . Bonus, M. (2011). Dissociation predicts poor response to dialectial behavioral therapy in female patients with Borderline Personality Disorder. Journal of Personality Disorders, 25(4), 432–447. doi:http://dx.doi.org/10.1521/pedi.2011.25.4.432 Kleindienst, N., Priebe, K., Gorg, N., Dyer, A., Steil, R., Lyssenko, L., . . . Bohus, M. (2016). State dissociation moderates response to dialectical behavior therapy for posttraumatic stress disorder in women with and without Borderline Personality Disorder. European Journal of Psychotraumatology, 7, 1-9. Klonsky, D. E., Victor, S. E., & Saffer, B. Y. (2014). Nonsuicidal self-injury: What we know and what we need to know. Canadian Journal of Psychiatry / La Revue canadienne de psychiatrie, 59(11), 565–568. 275 Kolla, N. J., Links, P. S., McMain, S., Streiner, D. L., Cardish, R., & Cook, M. (2009). Demonstrating adherence to guidelines for the treatment of patients with Borderline Personality Disorder. Canadian Journal of Psychiatry / La Revue canadienne de psychiatrie, 54(3), 181–189. Koons, C. R., Robins, C. J., Tweed, L. J., Lynch, T. R., Gonzalez, A. M., Morse, J. Q., . . . Bastian, L. A. (2001). Efficacy of Dialectical Behavior Therapy in women veterans with Borderline Personality Disorder. Behavior Therapy, 32, 371–390. Korner, A., Gerull, F., Meares, R., & Stevenson, J. (2006). Borderline Personality Disorder treated with the Conversational Model: A replication study. Comprehensive Psychiatry, 47, 406–411. Korner, A., Gerull, F., Meares, R., & Stevenson, J. (2008). The nothing that is something: Core dysphoria as the central feature of Borderline Personality Disorder implications for treatment. American Journal of Psychotherapy, 62(4), 377–394. Korner, A., & McLean, L. (2017). Conversational Model psychotherapy. Australasian Psychiatry, 25(3), 219–221. Kraemer, H. C. (2016). Messages for clinicians: moderators and mediators of treatment outcome in randomized clinical trials. American Journal of Psychiatry, 173(7), 672-679. Landes, S.J., Chalker, S.A., & Comtois, K. (2016). Predicting dropout in outpatient Dialectical Behaviour Therapy with patients with Borderline Personality Disorder receiving psychiatric disability. Borderline Personality Disorder and Emotion Dysregulation, 3(9), 1-8. Landes, S. J., Rodriguez, A. L., Smith, B. N., Matthieu, M. M., Trent, L. R., Kemp, J., & Thompson, C. (2017). Barriers, facilitators and benefits of implementation of 276 Dialectical Behaviour Therapy in routine care: Results from a national program evaluation survey in the Veterans Health Administration. Translational Behavioral Medicine, 7, 1–13. Lariviere, N., Couture, E., Blackburn, C., Carbonneau, M., Lacombe, C., Schinck, S.-A., . . . St-Cyr-Tribble, D. (2015). Recovery, as experienced by women with Borderline Personality Disorder. Psychiatric Quarterly, 86(4), 555–568. doi:http://dx.doi.org/10.1007/s11126-015-9350-x Le Boutillier, C., Chevalier, A., Lawrence, V., Leamy, M., Bird, V. J., Macpherson, R., . . . Slade, M. (2015). Staff understanding of recovery-orientated mental health practice: a systematic review and narrative synthesis. Implementation Science, 10(1), 87. Leichsenring, F., Abbass, A., Hilsenroth, M. J., Leweke, F., Luyten, P., Keefe, J. R., . . . Steinert, C. (2017). Biases in research: Risk factors for non-replicability in psychotherapy and pharmacotherapy research. Psychological Medicine, 47, 1000–1011. Leichsenring, F., Leibing, E., Kruse, J., New, A. S., & Leweke, F. (2011). Borderline Personality Disorder. Lancet, 377, 74–84. doi:10.1016/S0140-6736(10)61422-5 Levy, K. N., Clarkin, J. F., Yeomans, F. E., Scott, L. N., Wasserman, R. H., & Kernberg, O. F. (2006). The mechanisms of change in the treatment of Borderline Personality Disorder with transference focused psychotherapy. Journal of Clinical Psychology, 62(4), 481–501. doi:10.1002/jclp.20239 Lieb, K., Zanarini, M., Schmahl, C., Linehan, M. M., & Bohus, M. (2004). Borderline Personality Disorder. Lancet, 364, 453–461. 277 Linehan, M. M. (1993a). Cognitive-behavioral treatment of Borderline Personality Disorder. New York, NY: The Guilford Press. Linehan, M. M. (1993b). Global Rating Scale. Seattle, WA: University of Washington. Linehan, M. M. (1993c). Skills training manual for treating Borderline Personality Disorder. New York, NY: The Guilford Press. Linehan, M. M. (1998). An illustration of Dialectical Behavior Therapy. In Session: Psychotherapy in Practice, 4(2), 21–44. Linehan, M. M., Armstrong, H. E., Suarez, A., Allmon, D., & Heard, H. L. (1991). Cognitive-behavioural treatment of chronically parasuicidal borderline patients. Archives of General Psychiatry, 48, 1060–1064. Linehan, M. M., & Comtois, K. A. (1996). Suicide Attempt and Self-Injury Count (SASI- Count). Retrieved from http//depts.washington.edu.au/brtc/files/L- SASI%20Count.pdf Linehan, M. M., Comtois, K. A., Brown, M. Z., Heard, H. L., & Wagner, A. (2006a). Suicide Attempt Self-Injury Interview (SASII): Development, reliability, and validity of a scale to assess suicide attempts and intentional self-injury. Psychological Assessment, 18(3), 303–312. doi:10.1037/1040-3590.18.3.303 Linehan, M. M., Comtois, K. A., & Lungu, A. (2011). A brief or in depth interview in assessing self-injury: Balancing assessment effort and results. Seattle, WA. Linehan, M. M., Comtois, K. A., Murray, A., Brown, M. Z., Gallop, R. J., Heard, H. L., . . . Lindenboim, N. (2006b). Two-year randomized controlled trial and follow-up of Dialectical Behavior Therapy vs. therapy by experts for suicidal behaviors and Borderline Personality Disorder. Archives of General Psychiatry, 63, 757–766. 278 Linehan, M. M., Dimeff, L. A., Reynolds, S. K., Comtois, K. A., Welch, S. S., Heagerty, P., & Kivlahan, D. R. (2002). Dialectical Behavior Therapy versus Comprehensive Validation Therapy plus 12-step for the treatment of opioid dependent women meeting criteria for Borderline Personality Disorder. Drug and Alcohol Dependence, 67(1), 13–26. doi:http://dx.doi.org/10.1016/S0376- 8716%2802%2900011-X Linehan, M. M., Korslund, K. E., Harned, M. S., Gallop, R. J., Lungu, A., Neacsiu, A. D., . . . Murray-Gregory, A. M. (2015). Dialectical Behavior Therapy for high suicide risk in individuals with Borderline Personality Disorder: A randomized clinical trial and component analysis. JAMA Psychiatry, 72(5), 475–482. doi:http://dx.doi.org/10.1001/jamapsychiatry.2014.3039 Linehan, M. M., Schmidt III, H., Dimeff, L. A., Craft, C., Kanter, J. W., & Comtois, K. A. (1999). Dialectical Behavior Therapy for patients with Borderline Personality Disorder and drug-dependence. American Journal of Addictions, 8, 279–292. Linehan, M. M., & Wilks, C. R. (2015). The course and evolution of Dialectical Behavior Therapy. American Journal of Psychotherapy, 69(2), 97–110. Locher, C., Gaab, J., & Blease, C. (2018). When a Placebo is not a Placebo: Problems and Solutions to the Gold Standard in Psychotherapy Research. Frontiers in Psychology, 9, 1-3. Luborsky, L., Diguer, L., Seligman, D. A., Rosenthal, R., Krause, E. D., Johnson, S., . . . Schweizer, E. (1999). The researcher’s own therapy allegiances: A “wild card” in comparisons of treatment efficacy. Clinical Psychology: Science and Practice, 6(1), 95–106. 279 Luborsky, L., Rosenthal, R., Diguer, L., Andrusyna, T. P., Berman, J. S., Levitt, J. T., . . . Krause, E. D. (2002). The dodo bird verdict is alive and well—mostly. Clinical Psychology: Science and Practice, 9(1), 2-12. Man‐Son‐Hing, M., Laupacis, A., O’Rourke, K., Molnar, F. J., Mahon, J., Chan, K. B., & Wells, G. (2002). Determination of the clinical importance of study results. Journal of General Internal Medicine, 17(6), 469–476. Margison, F., & Shapiro, D. A. (1986). Hobson’s Conversational Model of psychotherapy—training and evaluation: Discussion paper. Journal of the Royal Society of Medicine, 79, 468–472. Marlowe, D. B., Kirby, K. C., Festinger, D. S., Husband, S. D., & Platt, J. J. (1997). Impact of comorbid personality disorder symptoms on outcomes of behavioural treatments for cocaine dependence. Journal of Nervous and Mental Disease, 185, 483–490. Martinez-Raga, J., Marshall, E. J., Keaney, F., BAll, D., & Strang, J. (2002). Unplanned versus planned discharges from in-patient alcohol detoxification: Retrospective analysis of 47 first–episode admissions. Alcohol and Alcoholism, 37, 277-281. doi:10.1093/alcalc/37.3.277 McMain, S., Guimond, T., Streiner, D. L., Cardish, R., & Links, P. S. (2012). Dialectical Behavior Therapy compared with General Psychiatric Management for Borderline Personality Disorder: Clinical outcomes and functioning over a 2- year follow-up. American Journal of Psychiatry, 169(6), 650–661. McMain, S., Newman, M. G., Segal, Z. V., & DeRubeis, R. J. (2015). Cognitive Behavioral Therapy: Current status and future research directions. Psychotherapy Research, 25(3), 321–329. 280 McMain, S. F., Links, P. S., Gnam, W. H., Guimond, T., Cardish, R. J., Korman, L., & Streiner, D. L. (2009). A randomized trial of Dialectical Behavior Therapy versus General Psychiatric Management for Borderline Personality Disorder. American Journal of Psychiatry, 166(12), 1365–1374. doi:http://dx.doi.org/10.1176/appi.ajp.2009.09010039 McMichael, A. J., Rolison, J. J., Boeri, M., Kane, J. P. M., O'Niell, F. A., & Kee, F. (2016). How do psychiatrists apply minimum clinically important difference to assess patient responses to treatment? MDM Policy and Practice, 1, 1-7. Meares, R. (2004). The Conversational Model: An outline. American Journal of Psychotherapy, 58(1), 51–66. Meares, R. (2005). The metaphor of play: Origin and breakdown of personal being. London, England: Routledge. Meares, R. (2012). Borderline Personality Disorder and the Conversational Model: A clinician’s manual. New York, NY: Norton. Meares, R., & Stevenson, J. (2000). Borderline Personality Disorder. Australian & New Zealand Journal of Psychiatry, 34(5), 869–871. Meares, R., Stevenson, J., & Comerford, A. (1999). Psychotherapy with borderline patients: I. A comparison between treated and untreated cohorts. Australian & New Zealand Journal of Psychiatry, 33(4), 467–472. Midgley, N., Ansaldo, F., & Target, M. (2013). The meaningful assessment of therapy outcomes: Incorporating a qualitative study into randomized controlled trial evaluating the treatment of adolescent depression. Psychotherapy (Chic), 1–10. doi:10.1037/a0034179 281 Mohr, D. C., Ho, J., Hart, T. L., Baron, K. G., Berendsen, M., Beckner, V., . . . Kinsinger, S. W. (2014). Control condition design and implementation features in controlled trials: a meta-analysis of trials evaluating psychotherapy for depression. Translational Behavioral Medicine, 4(4), 407-423. Morey, L. C., Benson, K. T., Busch, A. J., & Skodol, A. E. (2015). Personality disorders in DSM-5: Emerging research on the alternative model. Current psychiatry reports, 17(4), 1-9. Nadort, M., Arntz, A., Smit, J. H., Giesen-Bloo, J., Eikelenboom, M., Shiphoven, P., . . . Van Dyck, R. (2009). Implementation of outpatient schema therapy for Borderline Personality Disorder with versus without crisis support by the therapist outside office hours: A randomized trial. Behaviour Research Therapy, 47, 961–973. Nathan, P. E., Stuart, S. P., & Dolan, S. (2000). Research on psychotherapy efficacy and effectiveness: Between Scylla and Charybdis? Psychological Bulletin, 126(6), 964–981. National Collaborating Centre for Mental Health. (2009). Borderline Personality Disorder: Treatment and management. London: British Psychological Society. National Health and Medical Research Council. (2012). Clinical practice guideline for the management of Borderline Personality Disorder. Melbourne, Australia: Author. Neacsiu, A. D., Rizvi, S. L., & Linehan, M. M. (2010). Dialectical Behavior Therapy skills use as a mediator and outcome of treatment for Borderline Personality Disorder. Behaviour Research and Therapy, 48(9), 832–839. doi:http://dx.doi.org/10.1016/j.brat.2010.05.017 282 Ng, F. Y. Y., Bourke, M. E., & Grenyer, B. F. S. (2016). Recovery from Borderline Personality Disorder: A systematic review of the perspectives of consumers, clinicians, family and carers. doi:10.137/journal.pone.0160515 Norman, G. R., Sloan, J. A., & Wyrwich, K. W. (2003). Interpretation of changes in health-related quality of life: The remarkable university of half a standard deviation. Medical Care, 41(5), 582–592. O’Connell, B., & Dowling, M. (2014). Dialectical Behaviour Therapy (DBT) in the treatment of Borderline Personality Disorder. Journal of Psychiatric and Mental Health Nursing, 21(6), 518–525. Paret, C., Hoesterey, S., Kleindienst, N., & Schmahl, C. (2016). Associations of emotional arousal, dissociation and symptom severity with operant conditioning in Borderline Personality Disorder. Psychiatry Research, 244, 194– 201. doi:http://dx.doi.org/10.1016/j.psychres.2016.07.054 Paris, J. (2012). The outcome of Borderline Personality Disorder: Good for most but not all patients. American Journal of Psychiatry, 169(5), 445–446. doi:http://dx.doi.org/10.1176/appi.ajp.2012.12010092 Pasieczny, N., & Connor, J. (2011). The effectiveness of Dialectical Behaviour Therapy in routine public mental health settings: An Australian controlled trial. Behaviour Research and Therapy, 49(1), 4–10. doi:http://dx.doi.org/10.1016/j.brat.2010.09.006 Perepletchikova, F., Treat, T. A., & Kazdin, A. E. (2007). Treatment integrity in psychotherapy research: Analysis of the studies and examination of the associated factors. Journal of Consulting and Clinical Psychology, 75, 829–841. 283 Plener, P. L., Schumacher, T. S., Munz, L. M., & Groschwitz, R. C. (2015). The longitudinal course of non-suicidal self-injury and deliberate self-harm: A systematic review of the literature. Borderline Personality Disorder and Emotion Dysregulation, 2(1), 1-11. Pompili, M., Girardi, P., Ruberto, A., & Tatarelli, R. (2005). Suicide in Borderline Personality Disorder: A meta-analysis. Nordic Journal of Psychiatry, 59(5), 319– 324. Prendergast, N., & McCausland, J. (2007). Dialectic Behaviour Therapy: A 12-month collaborative program in a local community setting. Behaviour Change, 24(1), 25–35. Priebe, S., Bhatti, N., Barnicot, K., Bremner, S., Gaglia, A., Katsakou, C., . . . Zinkler, M. (2012). Effectiveness and cost-effectiveness of Dialectical Behaviour Therapy for self-harming patients with personality disorder: A pragmatic randomised controlled trial. Psychotherapy and Psychosomatics, 81(6), 356–365. Ribeiro, E., Ribeiro, A. P., Goncalves, M. M., Horvath, A. O., & Stiles, W. B. (2013). How collaboration in therapy becomes therapeutic: The therapeutic collaboration coding system. Psychology and Psychotherapy: Theory, Research and Practice, 86(3), 294–314. doi:10.1111/j.2044-8341.2012.02066 Rizvi, S. L. (2011). Treatment failure in Dialectical Behavior Therapy. Cognitive and Behavioral Practice, 18(3), 403–412. doi:http://dx.doi.org/10.1016/j.cbpra.2010.05.003 Rosenzweig, S. (1936). Some implicit common factors in diverse methods of psychotherapy. American journal of Orthopsychiatry, 6(3), 412-415. 284 Roy-Byrne, P. P., Sherbourne, C. D., Craske, M. G., Stein, M. B., Katon, W., Sullivan, G., . . . Bystritsky, A. (2003). Moving treatment research from clinical trials to the real world. Psychiatric Services, 54(3), 327–332. Schulz, K. F., Altman, D. G., & Moher, D. (2010). CONSORT 2010 statement: updated guidelines for reporting parallel group randomised trials. Trials, 11, 1-8. Shapiro, D. A., Barkham, M., Rees, A., Hardy, G., Reynolds, S., & Startup, M. (1994). Effects of treatment duration and severity of depression on the effectiveness of cognitive behavioural and psychodynamic-interpersonal psychotherapy. Journal of Consulting and Clinical Psychology, 62, 522–534. Shapiro, D. A., & Startup, M. (1992). Measuring therapist adherence in exploratory psychotherapy. Psychotherapy Research, 2(3), 193–203. Shapiro, D. A., & Startup, M. J. (1990). Raters’ manual for the Sheffield Psychotherapy Rating Scale (Memo no. 1154). Sheffield, England: University of Sheffield. Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in assessing rater- reliability. Psychological Bulletin, 86, 420–428. Silberschatz, G. (2017). Improving the yield of psychotherapy research. Psychotherapy Research, 27(1), 1-13. Slade, M. (2009). Personal recovery and mental illness: A guide for mental health professionals: Cambridge University Press. Spinhoven, P., Giesen-Bloo, J. H., Van Dyck, R., Kooiman, K., & Arntz, A. (2007). The therapeutic alliance in schema-focused therapy and transference-focused psychotherapy for Borderline Personality Disorder. Journal of Consulting and Clinical Psychology, 75(1), 104–115. doi:10.1037/0022-006x.75.1.104 285 Startup, M., & Shapiro, D. A. (1993). Therapist treatment fidelity in prescriptive vs. exploratory psychotherapy. British Journal of Clinical Psychology, 32, 443–456. Stevenson, J., Haliburn, J., & Halovic, S. (2015). Trauma, personality disorder and chronic depression: The role of the Conversational Model of psychodynamic psychotherapy in treatment resistant depression. Psychoanalytic Psychotherapy, 30, 23–41. Stevenson, J., & Meares, R. (1992). An outcome study of psychotherapy for patients with Borderline Personality Disorder. American Journal of Psychiatry, 149(3), 358–362. Stevenson, J., Meares, R., & D’Angelo, R. (2005). Five-year outcome of outpatient psychotherapy with borderline patients. Psychological Medicine, 35, 79–87. doi:10.1017/S0033291704002788 Stiglmayr, C., Stecher-Mohr, J., Wagner, T., Meibner, J., Spretz, D., Steffens, C., . . . Barbette, R. (2014). Effectiveness of Dialectic Behavioral Therapy in routine outpatient care: The Berlin Borderline Study. Borderline Personality Disorder and Emotional Dysregulation, 1 (1), 1-11. Stoffers, J. M., Völlm, B. A., Rücker, G., Timmer, A., Huband, N., & Lieb, K. (2012). Psychological therapies for people with Borderline Personality Disorder. Cochrane Database of Systematic Reviews, Issue 8, Art. No.: CD005652. Swenson, C. R. (2000). How can we account for DBT's widespread popularity? Clinical Psychology: Science and Practice, 7(1), 87-91. Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics. Boston, MA: Allyn and Bacon. 286 Tomko, R.L., Trull, T.J., Wood, P.K., & Sher, K.J. (2013). Characteristics of Borderline Personality Disorder in a community sample: Comorbidity, treatment utilisation and general functioning. Journal of Personality Disorders, 27, 1-17. Trull, T.J., Freeman, L.K., Vebares, T.J., Choate, A.M., Helle, A.C., & Wycoff, A.M. (2018). Borderline Personality Disorder and substance use disorders: An updated review. Borderline Personality Disorder and Emotion Dysregulation, 5, 1-12. Turner, R. M. (2000). Naturalistic evaluation of Dialectical Behavior Therapy-oriented treatment for Borderline Personality Disorder. Cognitive and Behavioral Practice, 7(4), 413–419. doi:http://dx.doi.org/10.1016/S1077- 7229%2800%2980052-8 Ulvenes, P. G., Berggraf, L., Hoffart, A., Stiles, T. C., Svartberg, M., McMullough, L., & Wampold, B. E. (2012). Different processes for different therapies: Therapist actions, therapeutic bond, and outcomes. Psychotherapy (Chic), 49(3), 291– 302. doi:10.1037/a0027895 van den Bosch, L. M., Koeter, M. W., Stijnen, T., Verheul, R., & van den Brink, W. (2005). Sustained efficacy of Dialectical Behaviour Therapy for Borderline Personality Disorder. Behaviour Research and Therapy, 43(9), 1231–1241. doi:http://dx.doi.org/10.1016/j.brat.2004.09.008 van Ijzendoorn, M. H., & Schuengal, C. (1996). The measurement of dissociation in normal and clinical populations: Meta-analytic validation of Dissociative Experiences Scale (DES). Clinical Psychology Review, 16, 365–382. Vaughan, B. R., & Ochoa, E. S. (2016). Optimizing adherence and outcomes in psychotherapy. American Journal of Psychiatry, 173(5), 444–445. 287 Verheul, R., van den Bosch, L. M., Koeter, M. W., de Ridder, M. A., Stijnen, T., & van den Brink, W. (2003). Dialectical Behaviour Therapy for women with Borderline Personality Disorder: 12-month, randomised clinical trial in The Netherlands. British Journal of Psychiatry, 182(2), 135–140. doi:http://dx.doi.org/10.1192/bjp.182.2.135 Walton, C. J., Goldman, B. B., Bendit, N., & Startup, M. (under review). Newcastle Adherence Scale for Conversational Model (NASCOM): Development and reliability of an adherence measure for Conversational Model psychotherapy. Waltz, J., Addis, M. E., Koerner, K., & Jacobson, N. S. (1993). Testing the integrity of a psychotherapy protocol: Assessment of adherence and competence. Journal of Consulting and Clinical Psychology, 61, 620–630. Webb, C. A., DeRubeis, R. J., Amsterdam, J. D., Shelton, R. C., Hollon, S. D., & Dimidjian, S. (2011). Two aspects of the therapeutic alliance: Differential relations with depressive symptom change. Journal of Consulting and Clinical Psychology, 79(3), 279–283. doi:10.1037/a0023252 Wetzelaer, P., Farrell, J., Evers, S. M., Jacob, G. A., Lee, C. W., Brand, O., . . . Arntz, A. (2014). Design of an international multicentre RCT on group schema therapy for Borderline Personality Disorder. BMC Psychiatry, 14, 319. Wilks, C. R., Korslund, K. E., Harned, M. S., & Linehan, M. M. (2016). Dialectical Behavior Therapy and domains of functioning over two years. Behaviour Research and Therapy, 77, 162–169. doi:http://dx.doi.org/10.1016/j.brat.2015.12.013 288 Williams, S. E., Hartstone, M. D., & Denson, L. A. (2010). Dialectical Behavioural Therapy and Borderline Personality Disorder: Effects on service utilisation and self-reported symptoms. Behaviour Change, 27(4), 251–264. Wise, E. A. (2004). Methods for analyzing psychotherapy outcomes: A review of clinical significance, reliable change and recommendations for future directions. Journal of Personality Assessment, 82(1), 50–59. Wnuk, S., McMain, S., Links, P. S., Habinski, L., Murray, J., & Guimond, T. (2013). Factors related to dropout from treatment in two outpatient treatments for Borderline Personality Disorder. Journal of Personality Disorders, 27(6), 716– 726. Yeomans, F. E., Hull, J., Delaney, J., & Clarkin, J. F. (2004). An instrument for rating therapist adherence to a psychodynamic treatment model. Weill Medical College, Cornell University. Zanarini, M. C., Frankenburg, F. R., DeLuca, C. J., Hennen, J., Khera, G. S., & Gunderson, J. G. (1998). The pain of being borderline: Dysphoric states specific to Borderline Personality Disorder. Harvard Review of Psychiatry, 6(4), 201–207. 289 Appendix A: Study Information Sheet Comparing Dialectical Behaviour Therapy and the Conversational Model in the Treatment of Borderline Personality Disorder: A pilot study. Information Sheet You are invited to take part in the research project identified above. My name is Dr Nick Bendit and I am a Psychiatrist at the Centre for Psychotherapy. This research is being conducted in association with Dr Carla Walton, Clinical Psychologist and Associate Professor Leslie Pollock, Clinical Director and Clinical Psychologist at The Centre for Psychotherapy, Hunter New England Mental Health. Why is this research being done? The purpose of this research is to compare two established treatments for Borderline Personality Disorder: Dialectical Behaviour Therapy (DBT) and the Conversational Model (CM). Previous research has shown both of these treatments to be helpful, however, no previous research has compared the therapies. The two treatments are quite different and it may be that each of them leads to improvements in different areas; hence, we are hoping that the research will assist us to determine that information. Who can participate in the Research? For this research, we are seeking people with a diagnosis of Borderline Personality Disorder, between the ages of 18 and 65, who have self-harmed on at least 3 occasions in the past year and are willing to engage in a psychotherapeutic treatment program. If you have a psychotic illness, an antisocial behaviour that poses a significant threat to staff, a developmental disability, or are unable to speak or read English, then unfortunately the study is not suitable for you. If you have a drug and/or alcohol dependency, the study is suitable for you but you will need to commit to attending all therapy sessions not intoxicated. If you have received prior treatment with DBT or the Conversational Model, the study will not be suitable for you. Participation in this study is entirely voluntary. Only people who give their informed consent will be included in the study. If you do not wish to participate in the study or should you withdraw from the study, this will not affect your care from the Centre for Psychotherapy or any other service in the Hunter New England Area Mental Health Services. You may withdraw from this study at any time without giving a reason and if you decide to withdraw from the study, you will have the option of withdrawing your data. The data will not be released to anyone else, except under the requirement of law. The researchers may withdraw a participant if it is considered in the participant’s best interest or if it is appropriate to do so for another reason. If this happens the researchers will explain why and advise you about any follow-up procedures or alternative arrangements as appropriate. 290 What would you be asked to do? If you choose to participate, you will be assigned, on a random basis, for example, like tossing a coin, to one of two treatment groups. You would be asked to attend the Centre for Psychotherapy twice weekly for therapy. Sessions will be conducted at the Centre for Psychotherapy at no charge to you. The two treatment groups are: A Dialectical Behaviour Therapy: If allocated to this therapy, you would attend weekly individual therapy for two months and then commence weekly individual and group therapy for 12 months. You will then have the option of continuing individual therapy for another three months. The group therapy will meet once each week for two and a half hours and focus on teaching skills in four main areas: being more effective in your relationships, learning to tolerate distress, learning ways to manage your emotions and developing the capacity to live in the present moment with awareness. The individual DBT therapy session will be for approximately one hour per week and the therapist will work collaboratively with you to identify problems in your life and then help you apply the skills that you have been learning in the group therapy to deal with these problems. B Conversational Model If allocated to this therapy, you will have individual weekly therapy for two months, twice a week individual therapy for twelve months, and then optional therapy for three months. The individual therapy will be for approximately one hour per session and will be nondirective. The focus will be on understanding your emotional experience and having this actively described back to you. If you agree to participate, you will be asked to participate in a brief interview and asked about your preference for different types of therapy. You will also be asked to complete a series of questionnaires taking approximately 2 hours with researchers at the Centre for Psychotherapy. The purpose of the questionnaires is to get a measure of how well you are doing at different points across the therapy. The questionnaires examine level of depression, amount and methods of self-harm, level of day-to-day awareness, difficulties with emotions and interpersonal relationships, services you have been involved with and your goals for treatment. You will be asked to complete the same questionnaires at the start of therapy, in the middle of the therapy, at the end of therapy and 12 months after completing therapy. Some of these questionnaires are self-report measures, whilst others involve you being asked a series of questions by an interviewer. What are the risks and benefits? The researchers expect that you will benefit from participating in this study, as you will receive an intensive psychosocial treatment, both of which have been shown to reduce symptoms associated with Borderline Personality Disorder. As the treatment is being given within a public psychiatric outpatient facility, there is no charge to you. There are no expected risks from participating in the treatment. We are aware that some of the experiences you have had and the exploration of these issues can be distressing to discuss and we would encourage you to discuss this with your clinician if you choose to participate, so that they can help you to manage this. Similarly, sometimes people find the data collection may be emotionally distressing. The interviewers are trained clinicians who will able to provide support during this process. 291 If reading this information sheet has caused recollections that you wish to talk about, but you do not wish to participate in this research, you may wish to contact Lifeline on 131114, the Psychiatric Emergency Centre on 1800 655 085, or your local Mental Health team of Hunter New England Mental Health. How will your privacy be protected? Your results will be treated respectfully and with confidentiality. Notes written following each session will be kept in your medical file and only available to relevant mental health staff. Questionnaire data will be kept at the Centre for Psychotherapy in a locked filing cabinet for 15 years. Your medical records may be accessed to obtain information about number of visits to mental health services and to measure occasions of self-harm. This personal information will be accessed, used & stored in accordance with Commonwealth Privacy Laws and the NSW Health Records and Information Privacy Act 2002. How will the information be used? The information collected will be used to determine the feasibility of a larger-scale research project. The data may be reported in scientific journals and presented at conferences. However, individuals will not be identifiable in any reports arising from this study. If you wish to have a summary of the study sent to you two years after completion of the project, please complete the relevant section of the consent form. What do you need to do if you wish to participate or get further information? Please read the information form carefully before you decide to participate. If you have any questions about the research, please contact Dr Nick Bendit, Dr Carla Walton or A/Prof Leslie Pollock at the Centre for Psychotherapy on 4924 6820. If you would like to participate, please complete the attached consent form and return it to Dr Nick Bendit at the Centre for Psychotherapy in the envelope provided within 3 weeks of receiving this invitation. If we have not heard from you in that time the receptionist from the Centre from Psychotherapy will phone you to see if you would like a researcher to contact you to answer any questions you may have. If you do not wish to participate in the study, you can refuse to participate when contacted by the receptionist. Please keep this information sheet for future reference. Thank you for considering this invitation. Dr Nick Bendit Dr Carla Walton A/Prof Leslie Pollock Consultant Psychiatrist Clinical Psychologist Clinical Director Complaints about this research: This project has been approved by the Hunter New England Human Research Ethics Committee of Hunter New England Health, Reference (Reference No. 06/12/13/5.11). Should you have concerns about your rights as a participant in this research, or you have a complaint about the manner in which the research is conducted, it may be given 292 to the researcher, or, if an independent person is preferred, to Dr Nicole Gerrand, Professional Officer (Research Ethics), Hunter New England Human Research Ethics Committee, Hunter New England Health, Locked Bag 1, New Lambton NSW 2305, telephone (02) 49214950, email [email protected]. 293 Appendix B: Study Consent Form Comparing Dialectical Behaviour Therapy and the Conversational Model in the Treatment of Borderline Personality Disorder: A pilot study. Consent Form Researchers: Dr Nick Bendit, Dr Carla Walton and A/Prof Leslie Pollock I agree to participate in the above research project and give my consent freely. I understand that the project will be conducted as described in the Information Sheet, a copy of which I have retained. I understand I can withdraw for the project at any time and do not have to give any reason for withdrawing. I consent to participate in the therapy and to completing the questionnaires listed in the Information sheet before, at the middle, at the end of therapy and 12 months following therapy. I understand that my personal information will be kept confidential to the researchers and appropriate Hunter New England Mental Health staff unless required otherwise by law. I understand that my therapy sessions will be audio-recorded for the purposes of supervision and to ensure that the therapist is providing the treatment consistently. I have had the opportunity to have questions answered to my satisfaction. I understand that my therapy sessions will be audio-recorded for research into understanding how therapy is effective. I do/do not (circle appropriate) want a summary of the study posted to me when the study is complete (I acknowledge that this may take up to two years). Print Name: Signature: Date: 294 Appendix 3: Rater’s Manual for the Newcastle Adherence Scale for the Conversational Model (NASCoM) RATERS' MANUAL For Newcastle Adherence Scale for the Conversational Model (NASCoM) An adaptation of the Sheffield Psychotherapy Rating Scale (SPRS) Bernard Goldman, Mike Startup, Carla Walton & Nick Bendit 295 Contents I. Introduction ...... 299 II. General comments about rating ...... 299 1. Rating therapist behaviours ...... 299 2. Rating therapist facilitation ...... 300 3. Prerequisite knowledge to rate the NASCoM ...... 300 4. Rating Extensiveness, not Quality ...... 300 5. Treatment of interventions that overlap other interventions ...... 301 6. Avoiding Haloed Ratings ...... 302 7. Use of Examples ...... 303 8. Making Distinctions ...... 303 9. Specific Instances Required for Rating ...... 304 10. Overlap Between Current Versus Prior Sessions ...... 304 III. Specific comments about items in the NASCoM ...... 304 1. Receptive listening ...... 304 2. Tentative style ...... 306 3. Language of mutuality...... 308 4. Patterns in relationships ...... 309 5. Focus on Client-Therapeutic relationship ...... 310 6. Emotional Attunement...... 312 7. Facilitate Awareness of Feelings ...... 315 8. Avoidance of Affect ...... 317 9. Acceptance of affect ...... 319 10. Explanatory Statements ...... 320 11. Metaphor ...... 322 296 12. Personal disclosure ...... 324 13. Limitations ...... 326 15. Frame Changes ...... 331 16. Warmth ...... 332 17. Rapport ...... 333 18. Agenda Setting ...... 334 19. Directiveness ...... 335 20. Providing reassurance ...... 337 21. Advice giving...... 338 22. Providing psychological techniques ...... 339 23. Psychoeducation ...... 340 24. Information gathering ...... 341 25. Homework Assigned or reviewed ...... 342 References ...... 342 Appendix A: Sheffield Psychotherapy Rating Scale ...... 343 Table 1: Items (17) which have been included in or adapted to the NASCoM ...... 343 Table 2: Items (41) in the SPRS not included in the NASCoM...... 345 Appendix A: The Newcastle Adherence Scale for Conversational Model (NASCoM)347 1. Receptive listening ...... 347 2. Tentative style ...... 347 3. Language of mutuality...... 347 4. Identifying Patterns in relationships ...... 347 5. Focus on Client-Therapeutic relationship ...... 347 6. Emotional Attunement...... 347 7. Awareness of Feelings ...... 348 297 8. Avoidance of Affect ...... 348 9. Acceptance of affect ...... 348 10. Explanatory Statements ...... 348 11. Metaphor ...... 348 12. Personal disclosure ...... 348 13. Limitations ...... 348 14. Disjunctions ...... 349 15. Frame Changes ...... 349 16. Warmth ...... 349 17. Rapport ...... 349 18. Agenda Setting ...... 349 19. Directiveness ...... 349 20. Providing reassurance ...... 349 21. Advice giving...... 349 22. Providing psychological techniques ...... 350 23. Psychoeducation ...... 350 24. Information gathering ...... 350 25. Homework assigned: ...... 350 298 I. Introduction This Raters' Manual is intended to accompany the Newcastle Adherence Scale for the Conversational Model (NASCoM) which is an adaptation of the Sheffield Psychotherapy Rating Scale (SPRS; Shapiro & Startup, 1990). The NASCoM adapts 17 items from the SPRS (12 exploratory items, two facilitative items, and three other items used as proscribed items in the NASCoM). The remaining 41 items from the SPRS were not included in the NASCoM. A detailed list of those items included, adapted, or excluded from the SPRS is found at Appendix A. By way of further background, the SPRS was in turn an adaptation of the Collaborative Study Psychotherapy Rating Scale, Form 6 (CSPRS-6) for use in the Second Sheffield Psychotherapy Project (see Shapiro, Barkham, Hardy, Morrison, 1990). The Manual attempts to explain comprehensively the basis for rating the items in the NASCoM which is found Appendix II. Thus the Manual contains information on every item in the scale. It is essential that the rater be familiar with the material in this Manual before making ratings on the NASCoM. The Manual begins with General Comments and Instructions to raters which are important in rating the NASCoM. The remainder of the Manual is organised according to Scale item number. Each item contains (where applicable): (1) The exact wording and format of the item as it appears in the Scale. (2) A restatement of or elaboration on the item's purpose. (3) Definitions of terms used in the item. (4) General guidelines for rating the item. (5) Examples of therapist behaviour which should and should not be considered in rating the item. (6) Specific rules for rating the item. (7) Important distinctions to be made between items. II. General comments about rating 1. Rating therapist behaviours. The NASCoM is designed to rate therapist behaviour. In rating the scale items it is important to distinguish the therapist behaviour (as much as possible) from the client behaviour in response to the therapist. This is not possible when rating items 16 and 17, Warmth and Rapport respectively, in which the client behaviour must be taken into account. For the remaining items, however, the rater should attempt to rate the therapist behaviour, not the client response to that behaviour. In rating therapist behaviour, the rater should consider what the therapist attempted to do, not whether these attempts met with success or failure. For example, in rating, Item 7, Facilitating awareness of feelings, the rater must determine how frequently the therapist facilitated the client’s awareness of feelings regardless of whether the client rejected the therapist’s attempts to name those feelings. 299 2. Rating therapist facilitation. One difficulty that arises in attempting to rate therapist behaviour (which is the goal of the NASCoM) is that sometimes the client initiates a behaviour which is measured in an item. Similarly, in other cases the client may actually engage in a behaviour being measured in an item with limited therapist involvement. An item should not necessarily be excluded in either of these circumstances. In these cases, ratings should reflect the degree to which the therapist facilitates the behaviour being measured. Here, facilitation refers to more than a passive acceptance on the part of the therapist of the client's behaviour. The therapist must actively engage with the client’s contribution. For example, in rating the facilitation of metaphor in therapy (Item 11), it is not for the therapist to create the metaphor for the client. The therapist is rated for what he or she does when the client offers symbolic content such as the metaphor. 3. Prerequisite knowledge to rate the NASCoM. Raters are not required to have special knowledge of the behaviours being measured by the NASCoM in order to rate the items. The NASCoM was specifically designed so that raters with no previous exposure to the Conversational Model could reliably and validly rate therapist behaviours which occur in CM. The Raters' Manual has been designed to provide all the background needed in order to rate the items. Nevertheless, it is important to note that the Conversational Model, unlike other models of therapy such as Cognitive Behaviour Therapy, provides no opportunities for didactic interventions. In fact, it proscribes such interventions. The Conversational Model promotes short therapeutic responses. The reason for parsimony is that one of the therapist’s goals include creating a space for the client to explore their inner life such as their feelings and imaginings. It follows that the raters need to be sensitive to any interventions as they are likely to be neither frequent nor wordy. It also follows that when using the NASCoM, the rater must be careful and conscientious in listening to and rating therapy sessions. It is ideal that each session is transcribed so that the interventions are categorised and substantiated. A transcription of the session should facilitate greater confidence in rating how much of the session involved a particular item. Rating is a complex task and requires the rater to be thoughtful and to exercise good judgment. 4. Rating Extensiveness, not Quality. The NASCoM is designed to measure the extent to which therapists engages in the behaviours being measured rather than the quality with which those behaviours are performed. Although extensiveness is not totally independent of the quality of therapist behaviour, the rater should not consider the quality of the therapist behaviour per se. 300 The following scoring system has been developed to assist in assigning ratings to signify the extensiveness of the interventions carried out in session. Score of 0 occurs when the therapist does not engage in any intervention in relation to a particular item. The score reflects the situation where the intervention is “not at all” present in session. Score of 1 is for interventions that amount to 5% of the therapist’s interventions in a session. Such a score is likely to be allocated in situations where the therapist occasionally uses an intervention, say one to three times in a session. Score of 2 reflects some use of interventions that take place more than 5% but less than 10% of the interventions used in the session. Score of 3 reflects some use of interventions that take place more than 10% but less than 20% of the interventions used in the session. Score of 4 reflects use of interventions that take place more than 20% but less than 25% of the interventions used in the session. Score of 5 reflects use of interventions that take place more than 25% but less than 30% of the interventions used in the session. A score of 6 reflects use of interventions that take place more than 30% of the interventions used in the session. To summarise Specific intervention Rating occurred % of all interventions Did not occur 0 1-5% 1 6-10% 2 11-20% 3 21-25% 4 26-30% 5 31 or more 6 5. Treatment of interventions that overlap other interventions. There are a number of interventions that clearly overlap. As mentioned above, process interventions (the “how to”) will take place in conjunction with other interventions. For example, it would be expected that the use of tentative language (item 2) will take place when an explanatory statement (item 10) is offered. In such cases, both items need to be rated. There are also a number of interventions that can be construed in different ways so as to trigger use of a number of items. For example, handling personal disclosure (item12) while 301 addressing the therapeutic client relationship (item 5). In such cases, unless specified otherwise in the particular items, both items can be rated. 6. Avoiding Haloed Ratings. The NASCoM was designed for the purpose of describing the therapist's behaviour in the session. In order to use it correctly, it is essential that the rater rates what is heard, not what she/he thinks ought to have occurred. The rater must be sure to apply the same standards for rating an item regardless of: a) what types of therapy the rater thinks she/he is rating. b) what other behaviours the therapist engaged in during the session. c) what ratings were given to other items. d) how skilled the rater believes the therapist to be in a particular modality. e) how much the rater likes the therapist. f) whether the rater thinks the behaviour being rated is a good or a bad thing to do. Example of rater halo resulting from rater's judgement of therapy modality (‘a’ above): The rater assumes that the item being rated is meant to measure an aspect of Modality A. This item might be rated higher than it should be as a result of the rater also assuming that the therapist was practising Modality A. Conversely, this item might be rated lower than it should be as a result of the rater assuming that the therapist was not practising Modality A. Example of rater halo resulting from consideration of other behaviours the therapist engaged in during the session (‘b’ above): In deciding what rating to assign to an item, the rater might erroneously base her/his ratings on behaviours which are similar to, or which are likely to covary with, the behaviours which are supposed to be considered in rating the item. Example of rater halo resulting from ratings given to other items (‘c’ above): In deciding what rating to assign to an item, the rater might erroneously base her/his ratings on ratings given to other items. This is likely to occur when the rater believes that the rating given to another item affects the rating given to the item currently being rated. For example, Example of rater halo resulting from rater's judgement of the therapist's level of skill (‘d’ above): The rater assumes the therapist is practising in Modality A. Furthermore, the rater assumes that the item being rated is meant to measure an aspect of Modality A. Based on these assumptions, the item might be rated lower than it should be if the rater judges that the therapist is not skilled in practising Modality A, and higher than it should be if the rater judges the therapist to be skilled in practising Modality A. Example of rater halo resulting from how much the rater likes the therapist (‘e’ above): 302 The two facilitative conditions, warmth and rapport, have limited criteria for determining rating. As it seems possible that with such items, positive regard for the therapist could contaminate the ratings, it is important to be mindful of avoiding the halo effect when rating these items. Example of rater halo resulting from rater's judgment of whether the behaviour is a good or bad thing to do (‘f’ above): The rater might assign a lower rating to an item than is warranted because he/she thinks the therapist is a good therapist and the behaviour being measured is undesirable. Similarly, the rater might assign a higher rating than is warranted because the rater believes the therapist is a good therapist and the behaviour being rated is desirable. 7. Use of Examples. For many of the items in this manual, we have given examples of therapeutic exchanges which provide guidelines for how to rate the therapist behaviour as adherent or non-adherent. Examples have been given when, in our experience with training raters, they have proven to be helpful. The examples in this manual are nevertheless only guidelines for how to categorise an item. Examples in the manual can occur in three different forms: a. A list of relevant aspects of the behaviour which should be considered in rating an item. b. A synopsis of a therapy exchange which should (or should not) result in a rating. c. A dialogue between the therapist and client which should (or should not) result in a rating. When a dialogue is given in an example, it is italicised and the letter "T" is used to indicate what the therapist said, and "C" is used to indicate what the client said. All names that appear in these examples are fictitious, as are most of the situations that are depicted. 8. Making Distinctions: Because the items vary in breadth of coverage, the same therapist behaviours which are rated under one item may also be rated appropriately under another item. Conversely, the rater is sometimes required to make fine distinctions between therapist behaviours which are similar yet should be rated distinctly. For example, in Item 12, Personal Disclosure, the rater rates the extent to which the therapist made appropriate use of self-disclosure, while in Item 13, Limitations, the rater rates, among other things, the extent to which the therapist addressed limitations such as making it clear to the client the limits to therapy around personal disclosures. Thus in rating these two items the rater must judge whether a disclosure is appropriate or inappropriate. Appropriate disclosures increase the rating on Disclosure because they further the shared exploration of what is happening between client and therapist. Refusals to make disclosures could also be rated against Limitations. 303 When possible, similar items have been placed near one another to help the rater make such distinctions. The rater should bear in mind the subtle differences between some items, and not use the same exact behaviour to substantiate ratings given to different items unless it is appropriate to do so. This Manual also contains an "Distinctions" section within the entry for some items. This section contains information regarding how the "target" item is similar and/or different from other "comparison" items. The "comparison" items contain a cross-reference to refer the rater to a discussion of how that item is similar to or different from the "target" item. The rater should not infer that the existence of these "Important Distinctions" means they are the only important distinctions that need to be made. All of the items are similar to or different from other items in important ways. Thus the rater should not rely on "Important Distinctions" to point out all of the important similarities or differences which exist. 9. Specific Instances Required for Rating. In order to rate an item greater than 0, the rater must hear a specific example of the therapist behaviour being rated. The rater should be careful not to rate behaviour as having occurred if she/he thinks it probably occurred but cannot think of a specific example. The starting point for rating each item in the scale is 0. With the exceptions of items relating to warmth and rapport, the rater should assign a rating of greater than 0 only if she/he hears examples of the behaviour specified in the items. In relation to warmth and rapport, it seems more appropriate to have a default around ‘4”. Depending on the level of warmth and rapport shown in session, the rating can be raised or lowered. While there is a default rating, the rater should still be able to justify their rating of these two items. 10. Overlap Between Current Versus Prior Sessions. Sometimes an issue that was discussed in an earlier session is implicitly or explicitly referred to in the session being rated. Discussions which took place in an earlier session should not be considered in determining a rating given to the current session. III. Specific comments about items in the NASCoM 1. Receptive listening: Did the therapist appear to allow silence to continue as a means of encouraging the client to talk? The purpose of this item is to measure the extent to which the therapist uses silence as a means of encouraging the client to talk (or continue talking). This item should be rated greater than '0’ only if the rater judges that the therapist used silence for the purpose of encouraging the client to continue talking. Distinct periods of silence on the therapist’s part can be counted as interventions and then rated as a percentage based on the total number of interventions rated in the session. Do not rate the therapist’s amount of silence relative to the client. Both can be silent much of the time or silent rarely. 304 Scoring also under warmth or rapport may be appropriate where the therapist appears to be deliberately withholding responses, or providing minimal encouragement, even when the client is clearly uncomfortable with the silence. 305 2. Tentative style: Did the therapist express his/her views concerning the patient's experiences and circumstances as tentative statements, open to correction, and inviting elaboration and feedback? This item assesses the 'how' of the therapist's talk. It is concerned with how the therapist conveys his or her response to what the client is saying. A tentative style has the purpose of conveying to the client that the therapist is not an expert on how the client feels, thinks, or acts; rather the therapist offers tentatively a possible understanding of what may be happening for the client. For example, the therapist may preface a statement with the following words, 'This is the way I see it now...but maybe I am wrong.' Tentative statements may well be definite (i.e. clearly 'owned' by the therapist) and often specific (i.e. referring to particular experiences and making quite detailed comments or observations concerning these). These qualities of definiteness and specificity follow from the therapist doing his or her best to be accurate. However, the therapist rating highly on this item acknowledges that he or she does not know which answers are right for the client. The therapist conveys his or her wish to be corrected, expressing a hope for communication which will lead on to dialogue, with an adjustment of misunderstanding. This wish is expressed in words, constructions and turns of phrase, as well as in the way they are spoken. The rater should watch for such indicators of tentativeness prefacing a therapist comment such as the following: 'maybe', 'it's almost as if', 'I'm not sure about this, but...', 'I wonder if...', ‘kind of’ ’like’. EXAMPLES While keeping in mind that the rater needs to take into account the extent that the therapist uses tentative style, the following examples illustrates a response typical of a therapist scoring '0' for a particular therapist statement. C: I think I've been spending more money lately because I just need to cheer myself up. T: That shows that you're obviously just going back to being a little girl who needs treats to prove someone loves her. Questioning of an interrogatory style (such as questions beginning 'why’ would clearly not be considered to be using a tentative style. Such questioning would not score any points under this item. -- See the following two interventions which would not be given a rating under this item. (a) T: Why did you do that? 306 (b) T: Wasn't that just another way of getting attention? The following statements are examples of therapist using tentative styles. (a) C: I think I've been spending more money lately because I just need to cheer myself up. T: I'm not sure about this, but maybe, in a way, when you feel miserable and alone, buying things is comforting. It's almost as if you're feeling again like that little girl who needs treats, as if having things for yourself helps you feel comforted, and maybe loved. (b) C: That's just it, you see, you don't seem to want me to get over it. T: You feel that I don't want you to get over it. That feels pretty important to me. Let me try and put into words what may be happening here. It feels to me as if you're disappointed and so maybe a bit angry with me, as if you feel I'm not really with you, not really on your side. Maybe, that when I try to help us look at what's happening when you make such an effort to cheer yourself up, for you it's almost an attack. 307 3. Language of mutuality: Did the therapist use the language of shared endeavour (avoid the use of “you” and “I” and instead, used impersonal language or “we”)? This item measures how the therapist conveys his or her participation in the therapeutic conversation. The Conversational Model encourages the therapist to use language in a way that promotes an atmosphere where the therapist and client are looking at the issues together in contrast to a therapeutic atmosphere of the therapist interviewing the client. The use of the impersonal tense or the use of “we” are considered the most important means of creating a language of mutuality as the words “you” and “I” are not used in such circumstances. However, it is important to appreciate that there will be times when it will be preferable to use “you” especially in summary or explanatory statements; otherwise the language of the therapist may sound clumsy. It follows that when the therapist uses “you” this does not occasion a form of discounting of the overall rating. There are times when the therapist may appear to be simply excising the word “you” from a statement but where the word “you” is fully implied. The rater should treat these situations the same as if “you” were not used. EXAMPLES The following example would not contribute to a positive rating under this item. In this example, the therapist has located the problem only within the client: T: This is a big part of what happens when you try to get close to someone, you both want it and yet, in a way, you fear it too. The following intervention is an example of using impersonal language. C: It's always hard for me to get started each time I come. I don't know why, but I just feel awkward talking to you about the kind of thing that came up last week. T: Yes, it is difficult to discuss those sort of issues. ------ 308 4. Patterns in relationships: Did the therapist draw parallels or point out patterns in two or more of the client's relationships? The theoretical underpinnings of CM include the understanding that we learn in relationships. We are social beings who cannot learn without another human being. One of the most important types of learning we do is to socialise with other humans. For the person with BPD, there is some appreciation that early social environments provided a person with BPD with learning which is often dysfunctional in adulthood. If the patterns relate to the client-therapeutic relationship, they need to be scored under item 5. However, if the therapist makes connections with not only the client-therapist relationship but another client relationship, the rater can consider the intervention under this item as well as item five. EXAMPLES The following are examples in which the therapist pointed out a pattern in the client's successive relationships (example a) and a parallel between two of the client's current relationships (examples b and c). Example (a) T: It seems easy for you, both in this present relationship and in past significant other relationships, to bend to meet the other person's needs and to neglect yourself. Example (b) C: I really get angry when my friend starts telling me what to do. Whenever we get together she has advice for me on how I ought to do this or how I ought to do that. T: That sounds similar to the reaction you have when your boyfriend gives you advice. It might be helpful for us to find out what that's about. Example (c) C: I stormed out of the flat after my argument with Joe and did not come back for a long time. T: This sounds similar to an incident with your mother that we spoke about a few sessions ago. 309 5. Focus on Client-Therapeutic relationship: Did the therapist address the client's present feelings around the therapeutic relationship or make links with the client’s present feelings about the therapeutic relationship with feelings in other contexts and at other times? This item measures the therapist's exploration of the client’s feelings about the therapeutic relationship by making links (a) between events within therapy at different times: perhaps during one interview, perhaps relating what is happening in the present session to what happened in previous sessions; or (b) between patterns in the present therapeutic conversation and those in other areas of life (especially ways in which relationships are defective and distorted). Through use of observed recurrent patterns in the client's experiences and making links with the client-therapist relationship, the therapist helps the client to make sense of experiences by helping the client create greater 'wholes' and thus to counter fragmentation and loss of integration. Any attempts by the therapist to address what is happening in the client- therapist relationship are also scored under this item as it is implicit that they point to other relationships or the therapeutic relationship in the past or previous sessions respectively. Examples The following example is considered an intervention under this item because the therapist made a link between events within therapy in different sessions. C: So you are going away again. T: Yes, I recall that it was hard for you when I was away last time. The next two examples show how the therapist made links between patterns in the therapeutic relationship and those in other areas of life. C: I stormed out of the flat after my argument with Joe and did not come back for a long time. T: I am wondering whether there is some concern that this might happen here. ______ C I really find it difficult when he asks me about my movements over the next few days. He seems to want to tie me down, and that annoys me. I just wish he wouldn't hassle me so much, and just let me decide for myself how to allocate my time. T: I wonder if you feel I'm trying to tie you down too, when I want to know how you're feeling about things. ------The next example is directly concerned with the resolution of misunderstanding in the therapeutic relationship. 310 C: You're probably right, I shouldn't be so bad tempered with Joe, because he doesn't realise what he's doing. T: Mhm....It sounds as if you feel I'm finding fault with you for feeling angry with Joe, so you feel criticized or maybe even attacked by me as well as by Joe. That must feel very painful. C: Yes, well, it did seem that you were telling me off for getting worked up about something minor, somehow. 311 6. Emotional Attunement: Did the therapist attune to the emotional cues or other words offered by the client by using any of the following three micro-skills: coupling, amplification, and representation? This item is intended to measure the extent to which the therapist conveys to the client that she/he has an intimate understanding of the client's experiences and feelings and their meaning to the client by using coupling amplification. and/or representation. Each micro skill with examples are set out below. Coupling Coupling is a way of attuning to the emotional life of the client by acknowledging the emotional tone used by the client or implicit in the client’s story. Coupling is at the heart of this model, and should be frequently used in every session. There are three ways that the therapist can respond to the emotional cues of the client in this item. Firstly, the therapist can respond to the emotional tone of the client’s voice, either by using a simple empathic statement/question (“you sound upset?”). Secondly, the therapist can make a more complex response to the content of what the patient is saying, but including some attempt to understand the feeling (For example, the therapist may observe a feeling of confusion in the following response, “It seems like when your mother said that, you just didn't know what to do next.”). Thirdly, the therapist can “couple” by way of picking up affect-laden words or images, and asking for elaboration. EXAMPLES OF COUPLING C: I get really fed up with Joe every time he brings up the subject of money. He has no idea how much this annoys me. T: That must be frustrating. Joe does not have a clue. It is annoying that he does not know you better. The following is an example of picking up affect-laden words: C: I just don't know what was happening that day, the whole thing was a horror, and yet nothing was out of the ordinary T: Horror? Amplification: Did the therapist offer statements that enlarge or add to the feelings that the client is talking about or experiencing? Amplification is another micro skill that allows the therapist to tune into what the client may be feeling. It is more than coupling or mirroring the client’s feelings. Rather, it offers a space or invitation for the client to more fully experience and talk about their feelings. Amplification enlarges or adds to the feelings that client is talking about or is experiencing. 312 EXAMPLE OF AMPLIFICATION C: I get really fed up with Joe every time he brings up the subject of money. He has no idea how much this annoys me. T: It is infuriating. Representation Representation aims to achieve a dialogue between therapist and client with increasing mutual understanding, in which the desire to understand is communicated. This involves more than simple repetition or reflection of the client's message. Representation requires that something more comes from the therapist's own perspective on the experiences reported by the client. Representation is acknowledging feelings, often conflicting feelings, and representing a larger internal picture. It is more than amplification of feelings but a re-presenting of what has been said in a way that extends rather than merely mirroring the client’s feelings. EXAMPLES OF REPRESENTATION Example (a) C: I get really fed up with Joe every time he brings up the subject of money. He has no idea how much this annoys me. T: You feel really mad with Joe, and maybe it's not just for bringing up the subject of money, but also for not understanding how much he upsets you by doing it. C: Yes, he's so clueless about how I feel, I suppose I reckon he couldn't care less.... Example (b) C: I just don't know what was happening that day, the whole thing was a horror, and yet nothing was out of the ordinary T: Horror? C: Yet, everyone was going off like firecrackers, my whole family, but it was all strangely familiar. T: So the weird thing seems to be that there was chaos, but it was normal too? DISTINCTION Note that explanatory statements are treated under item 10. The rater will be able to distinguish a representation from explanatory and summary statements on the basis that the purpose of representation is to empathically attune with the client. Explanatory statements are more oriented to providing an explanation 313 about behaviours or feelings. Finally, summary statements often involve a few sentences which have the purpose of summarising therapy rather than offering empathic understanding or explanation. Summary statements are not rated 314 7. Facilitate Awareness of Feelings: Did the therapist encourage the client to consider unspoken feelings of which the client may be unaware. For a therapist to receive a rating against this item there must not be a direct clue as to what the client is feeling. The therapist is rated under this item for their capacity to speculate about the client’s feelings. Therapist’s speculation may be informed by the emotional tone of the client, non-verbal cues, and feelings engendered in the therapist. The therapist can facilitate such awareness by identifying verbal and non-verbal cues supplied by the client such as posture, gestures, facial expression, and tone of voice. No doubt there are degrees of explicitness with which a therapist might base his/her facilitation of awareness. Another means of facilitating awareness of feelings is wondering aloud about possible feelings and in so doing help the client to recognise and label emotions that the client is unaware of or the client is aware of but is not expressing. EXAMPLES The following two examples of interventions are explicitly cue-based Example (a) T: As you tell me about what's been happening at home this week, you're looking like you're ready to cry. It sounds like things are pretty desperate and it is bringing up a lot of sadness. Example (b) T: As we talk about this, I've noticed you clenching your fists and blinking your eyes a lot. I wonder whether there is some anger about this. C: Well, let's see....I suppose I'm feeling angry that she'd do this to me. T: Your eyes were filling up, with feelings of anger, or maybe... C: Well, no....I reckon I feel pretty sad, too...I'm angry she'd do this, but sad that it is happening. I suppose I have lots of feelings. ______ The following three examples show how awareness can be facilitated by wondering aloud about the client’s feelings Example (a) T: It seems like the story is not important to you. You sound matter of fact when you relate that story to me. But I can't help wondering about your feelings towards your boss about her doing that to you. C: Now that you mention it, I suppose I am pretty miffed about it. But I was more focused at the moment on the fact that I didn't get a promotion rather than how I felt about my boss as a result. 315 Example (b) C: John said I was wrong to react like I did, but I just couldn't help it, somehow. I just had to tell them that I wasn't in a position to help out yet again, that they'd been taking it for granted that I would, but that it just wasn't possible. It happened again on Tuesday. Bill said that his section were short-staffed, and could I let them have an extra person for the day, despite their already having had extra help the previous day, which is much more than any section would normally ever have in a week. T: But I wonder what you're feeling, inside, as you tell me about all these demands people are making of you at work. Example (c) The therapist may be feeling something which is not obvious to the client. The therapist may tentatively say to the client who is talking about sadness T. I know you’re pretty upset about this but I wonder whether there is also anger? ------ Distinctions Items of coupling, amplification and representation increase the awareness of the client’s feelings. However, the primary goal of these items is to convey to the client that the therapist is interested in understanding and/or understands the client’s feeling life. Unlike emotional attunement, when the therapist facilitates awareness of feelings, the therapist is more speculative than empathetic. The therapist’s interventions in this item are more tentative than coupling, amplification, or representation. It should not be a surprise to the rater that the client may reject the therapist’s efforts to bring about awareness. The client’s rejection of the therapist’s comment about the client’s possible rejection does not limit the ratings for this item, rather, such rejections are often a sign that the therapist is attempting to facilitate awareness of feelings rather than intervene with emotional attunement. When rating this item, the rater needs to distinguish the client’s avoidance of affect (see item 8). Addressing avoidance relates to the therapist directly referring to perceived avoidances by the client. 316 8. Avoidance of Affect: Did the therapist assist the client to address any avoidances experienced by the client? This item is concerned with the therapist's description to the client of his/her tendencies to avoid painful or problematic thoughts and feelings. Any direct confrontation is not promoted under the CM. Rather, the therapist intervention should merely describe the avoidance for the client so as to facilitate the client to then proceed and process the avoided material. Typically, the therapist's message is not 'No, it isn't as you say but, instead, it is like this and it's time you stopped avoiding facing up to it', but rather, 'Yes, I hear what you say, but there may be more to it as well, and I'd like you to look at that possibility too.' Note that, as in the second example below, the therapist's efforts to facilitate the client to acknowledge his or her avoidance need not be successful. On the other hand, the client might begin to explore without the therapist needing to direct him/her to do so. This could also be rated under this item providing the therapist can be seen to have facilitated this exploration by the client. EXAMPLES Example (a) C: There's nothing difficult about this situation at all, really, I just don't like being with him, so I keep out of his way. That's all there is to it as far as I'm concerned. T: Yes, sure, you don't like being with him. But I can't help feeling there's something important here in this not liking being with him. C: Well, yes, his behaviour is so unreasonable, I just have to keep my distance from him as far as possible. T: Maybe that makes you comfortable because it protects you from something inside. C: There's just so much tension, so much pressure when I'm with him, I don't know why. T: Sort of feeling tense and pressurised, tense inside Example (b) T: Your boyfriend moved out for good this week without you expecting it. C: Yes, I mean I knew we were having some problems but I didn't expect him to just leave and say he wasn't coming back. T: So I suppose that's affected what you've been doing since he left. C: Yes, I've been working a lot, putting in extra time to stop me thinking about it because when I do, I start to feel so sad. 317 T: So, you've been trying to avoid having time to think about it and feel sad about it. C: Yes, that's what I've been doing. I feel like a nervous wreck, too. 318 9. Acceptance of affect: Did the therapist encourage the client to accept feelings of which the client is aware but which are painful or uncomfortable? The purpose of this item is to measure the extent to which the therapist discusses the client's acceptance of feelings which she/he acknowledges experiencing but has difficulty accepting. EXAMPLES The following are examples of ways in which the therapist might help the client to accept feelings which are painful or uncomfortable: (a) The therapist helped the client to understand and feel comfortable about feeling anger toward a loved one who was engaging in behaviours which the client thought were self-defeating. It is important to distinguish acknowledgment of affect from providing reassurance. Providing reassurance is not part of CM. T: No wonder you feel anger. [It would not be adherent to the CM if the said “It is OK to feel angry.” Such an intervention would be rated on the proscribed item relating to providing reassurance (see item 20)] ------ (b) The therapist helped the client understand why she/he was experiencing feelings of relief and happiness about the death of a parent who had been a burden to her/him. T: It makes sense you feel relieved that your father died. ------ (c) The therapist attempted to help the client to be comfortable with feeling happy in the face of the client's concern that she will be setting herself up for a big "let down" if she does feel happy. T: It is not surprising you feel confused. ------ DISTINCTIONS This item differs from Awareness of Feelings and Avoidance of Affect as this item deals with feelings that the client is aware of but does not want to experience, or wishes to reject. 319 10. Explanatory Statements: Did the therapist introduce explanatory statements which offer possible reasons (which may or may not be 'causes') for the client's behaviour and experiences, particularly in respect of disturbances in relationships within and outside therapy? It is not expected that explanatory statements will be used frequently in the Conversational Model especially in the early phase of therapy (first 16 weeks). It is desirable that the client should contribute some or all of the explanation themselves, so that client contributions to this are 'credited' to the therapist in making the rating, provided that there is evidence that the therapist has contributed to the client's arrival at the explanation. The rater should include interventions under this item where the therapist summarises what the client has said for the purposes of making explanatory links. Where a summary of the client’s issues is provided without making explanatory links, then the rater is not to include them under this item (or any other item for that matter). EXAMPLES To aid rating, examples four ways in which therapists may attempt to explain the client's current difficulties are set out below: (a) Conflicts between various personality tendencies: T: So although you would really like the security of a relationship and feel compelled by that, you're scared and find yourself running from the possibility of an intimate relationship ______(b) Events or motives rooted in the client's past: T: It seems to me that your indecisiveness regarding the possibility of separating from your wife points at that deeper conflict we've talked about -- the trouble you had earlier in your life around leaving your home and your mother. T: It seems that an intimate relationship is attractive but bearing in mind your childhood experiences a close relationship seems frightening. ______(c) Basic personality tendencies which influence the client's reactions to the therapist (transference): T: I wonder if you also feel powerless to resist the demands you perceive me making, just as you felt powerless to resist your father’s demands. ______(d) Motives or personality tendencies which serve to reduce anxiety or avoid discomfort: T: It looks like there's a pattern here. Whenever you're in a situation that could potentially result in your being asked out, you either get too sick to stay or you see yourself become very loud and abrasive. It seems like you are terrified of being involved with women. 320 DISTINCTIONS It can be difficult to distinguish explanatory statements from representations which form part of emotional attunement. Use of representation is for the purposes of empathic attunement. Explanatory statements are not oriented to empathic attunement. Rather, the use of explanatory statement is for the purpose of providing insight to the client about connections amongst the client’s behaviours, feelings, or relationships. A statement should not be construed as an explanatory statement unless it is clear that causal relationships are being made by the therapist or by the client with the help of the therapist. It is possible that explanatory statements could include a representation. When a statement can be broken down in sentences where one sentence can be attributed to representation and another to an explanatory statement, then each item can be rated separately. 321 11. Metaphor: Did the therapist encourage and elaborate the client's use of metaphor? The therapist is rated on occasions when he or she encourages the client to use metaphoric communication, elaborates or builds upon metaphors introduced by the client. The purpose of this intervention is for the therapist to make greater 'wholes' of the client's experience, and to heighten or intensify the client's experiencing and expression of feelings. There is some expectation that the client will not offer metaphors in the early part of therapy. The client’s use of metaphor in some way reflects developments in therapy where the client’s symbolic content and language develops as therapy unfolds. The rater should be alert to the possible use of metaphor by the client as an adornment rather than as direct and vivid communication. It may therefore sometimes be appropriate for the therapist not to respond to excessively ornate or elaborate metaphoric communication that possibly is serving a defensive function on the client’s part. The rater should also be alert not to rate use of metaphors simply because metaphors are used. Many metaphors are conventional figures of speech, such as words like “journey or road”. Rather, the rater needs to focus on how the therapist works with the client’s use of symbolic language. EXAMPLES The following examples show the therapist promoting the client's use of metaphor, teasing out and elaborating metaphoric content of which the client may be scarcely aware: Example (a) C It seems such a heavy burden of responsibility when I have to chair the meeting on a Friday. It's too much for me to cope with. T: The weight of that burden feels really overwhelming, maybe to the point where it was crushing. C: I find it very difficult to carry out the responsibilities of my job. There is very little guidance laid down. I'm all at sea with the work, especially at this time of year. T: No guidance, all at sea. It seems like you're at the mercy of the waves. Example (b) C: There's really not much going on in my life at the moment. It feels like an empty, quiet time. T: Empty, quiet, a void of stillness, that's a powerful image. C: I do feel that, a chasm, I suppose, opening up between me and everything or everyone around me. 322 T: The vastness of the distance between you.... 323 12. Personal disclosure: Did the therapist disclose or respond to client’s personal questions about the therapist in way that advanced a shared understanding of events and processes in the therapy? Appropriate disclosures to clients’ questions are those that advance a shared understanding of the client's interpersonal behaviour and experiences. Disclosures that serve the therapist's personal needs rather than those of the client, or disclosures that respond unreflectively to the client's demands or other communications, are not rated at all. An important therapeutic goal when making a personal disclosure is to accept the validity of the client’s question yet respond in a way that re-orients the client to what the client is experiencing in the therapeutic relationship. If the response also addresses limitations in therapy, the response can also be rated under Item 13 (Limitations in therapy). If the response addresses the client-therapist relationship, it can also be rated under item 5 (Linking patterns in the client-therapist relationship). EXAMPLES The following disclosure would not be rated under this item because it is an inappropriate response as it responds without reflecting on the client's demands: C: There's something I've been wanting to ask you. I mean, do you like me? T But of course I do. You're a very attractive person. ______The following example would also not be considered an appropriate disclosure because the therapist refuses to acknowledge his experience in relation to the client: C: There's something I've been wanting to ask you. I mean, do you like me? T: I don't think I should answer a question like that. ______The following example would be rated under this item. C: There's something I've been wanting to ask you. I mean, do you like me? T: You've been wondering about this for some time now, whether or not I really like you. Maybe that tells us something important about what's been happening between us. [Notice no disclosure at this point of the exchange.] C: Well, it's quite simple really, I just wondered, that's all. I mean, sometimes I think you're just doing your job, and don't seem that interested in me as an individual. 324 T: So there's something about how I come across to you that makes you feel, in a way, that I don't care for you as a person. [Again, note that there is no personal disclosure.] C: Mhmm, it seems kind of businesslike and a bit routine. T: Yes, perhaps my concern, which I do feel, to do things right and not let you down in terms of doing my job as I should, maybe that does come across as distant, or uncaring. [The therapist has to acknowledge his or her feelings when confronted yet use such acknowledgements to stay in tune with what the client may be feeling.] ______ Sometimes personal disclosure is useful to reduce client anxiety and keep the focus on the therapy. It may be appropriate to answer the client’s question directly so that the focus returns to therapy. C: Do you believe in Jesus? T: Can you tell me what is worrying you to ask this question? One way of returning the focus to the client is asking the client how they feel about the answer. An alternative response: C: Do you believe in Jesus? T: answers question and then asks “how do you feel about my answer to your question?” Distinctions A number of these examples could also be rated under linking patterns in the therapeutic relationship in situations where the therapist uses the client’s requests for personal information to encourage reflection on the therapist-client relationship. 325 13. Limitations: Did the therapist promote the client's exploration of feelings concerning the limits to therapy, and boundary, loss and internalisation issues related to termination? This item assesses the extent to which the therapist works to enable the client to deal with issues arising from the necessary limitations of therapy. These include the fixed limits to the number and length of treatment sessions, restrictions on the behaviour of both participants arising from their roles as therapist and client, and imperfections in or limitations to the therapist's ability to understand and help the client. Issues arising from these limits may include difficulties in maintaining boundaries or adhering to 'rules' governing relationships, painful feelings of loss or abandonment, or angry resentment at the therapist's withholding of the 'gifts' of time, personal disclosures, &c. Internalisation issues concern the ability or otherwise of the client to retain a positive sense of the therapy or therapist whilst yet acknowledging the limitations to what therapy or the therapist has offered. EXAMPLES The following intervention would not receive a score under this item as the therapist mentions to the client he is going on holidays (an example of a limit to therapy) without drawing out the client's reactions to these: T As you know, this is our last meeting before I'm away on holiday for three weeks. ______ The following intervention would be rated against this item as the therapist identifies a therapeutic boundary but also validates the client. C: And another thing I wanted to tell you about today is that my wife and I had a bit of an argument over the children on Tuesday, because she wanted to let them stay out later because it was school holiday time, but I thought it wasn't a good idea. T: That certainly sounds like something we should talk about in a future session, but I'm afraid there isn't time to do that today as we must finish in a minute or so. ______The following example would also be rated under this item as it reminds the client of a boundary (namely, the therapist having holidays) and invites the client to explore how he or she feels about the therapist going on holidays. T: As you know, this is our last meeting before I'm away on holiday for three weeks. I was wondering what this is like for you, I was wondering whether it may be harder for you to talk about difficult things today. ______ 326 The following example shows the therapist encouraging far-reaching exploration of a range of feelings in relation to limits, boundaries or termination. Very often, such an exploration addresses ambivalence or mixed feelings. C I have been wondering, lately, how far I can take this process of standing up for myself at work, and not just going along with other people's expectations of me. T: Yes, and maybe that's particularly important just now, as we're coming to the end of our meetings. C: And most of the time I don't like to dwell on the fact that we've nearly finished. OK I do feel different, but there're still lots of situations that I find hard. T: It sounds like you have mixed feelings about coming to the end of therapy. ______ In the next example, the therapist makes a worthwhile attempt to establish the boundary between irrelevant or unhelpful disclosures and appropriate sharing of feelings aroused in the therapeutic relationship itself. The rater could also rate under item 12 (Personal Disclosures) as the therapist addresses both limitations and personal disclosure. C: You know, I don't find it easy to keep on telling you all these things about myself, when I know so little about you. It seems so one-sided, and I do wonder how much it is safe to trust you, when you don't seem to want me to know you at all in any way. T: It sounds frustrating. You share so much of yourself yet you know so little about me. C: Yes, you expect me to tell you about very painful and private things, but there's no reciprocation. T: It may feel safer if I spoke about myself but it would be getting away from you. ______ There may be occasions to respond to the client and provide some self- disclosure so as to settle the client’s anxiety so that the therapeutic work can continue. For example, the therapist could say: T: Sure, there are lots of things I don't tell you, because I don't think it'd really be helpful to you if I did. But let's look back over what happened earlier today, when you told me about the rows you've been having at home and how pent up and angry you felt. I certainly felt the force of your anger, and tried to put into words how powerful it felt to me, and how I could really feel how frightening that could be for you. 327 In the above example, the therapist dealt with boundary issues and so should be rated under this item. Depending on the context of this intervention, it may be appropriate for the rater to rate under item 12 (Personal disclosure). The main reason for rating under both items would be when the therapist has spent time both discussing boundaries and responding to client’s questions about the therapist. The rater could also have rated under item five as it dealt with the therapeutic relationship. ------ In the next example, the therapist deals with the client's request for flexibility in the length of treatment sessions. T: Sure, it's very frustrating when we can't deal with something that’s so important to you. I’m wondering if we can talk about it next time. C: So why can't you let me stay and tell you about it now? T: I know this is important to you and it would be good to talk about it in next session. ______ The final example illustrates a failure to acknowledge limitations to the therapeutic relationship, by responding to the client's demands with an inappropriate disclosure that violates the boundary rules. This intervention would not be rated under this item. C: I know it's half past, but it's really important to me to tell you about this now, while it's fresh in my mind, rather than have to wait until next week. It's not much good if I can't tell you about things when they're happening, is it? T I wish I could stay to hear about this, but I must leave on time today as I have to take my wife to her hospital appointment, and she'll worry if I'm late. The therapist would have been rated if he had said: T: I know that it is really important but we can’t deal with it in a couple of minutes. It is too important. We will need to deal with it next time. 328 14. Disjunctions: Did the therapist address any disjunctions in therapy? Generally, disjunctions become apparent in the following circumstances: • when the client’s response seems muted or the client becomes silent., • when the client’s mood changes quickly, • when the client says something intellectual, or • when the client appears to dissociate. Disjunctions may be triggered by therapist behaviour such as incorrectly names the client’s feelings, or overlooking important information provided by the client, such as the death of a parent. Sometimes disjunctions are only tangentially related to the therapist. Whatever the cause of the disjunction, the therapist needs to address it in a timely manner. It follows that one of the skills in CM is to identify when a disjunction occurs. EXAMPLES (a) C: [Client suddenly becomes silent.] T: I am wondering what has happened in therapy in the last minute or so. Something has changed. I wonder what is happening. (b) T: What went wrong? C: Kylie, my daughter. T: Of course, I have forgotten Kylie [who was the client’s still born child.] C: It felt like you punched me. T: It must have felt hard. I am sorry. In the next example, the client complains of the therapist's failure to understand what she has been saying, and the therapist makes a constructive effort to deal with it in terms of limitations: (c) C: But you don't seem to understand how hard it is for me to work this out with her. You seem to think it's a simple matter of telling her my position. But that's just not possible. I sometimes despair of getting much real sympathy or understanding from you. 329 T: I am sorry. I can hear that I am not understanding. I have not appreciated the real difficulty of dealing with her. ______ In the next example, the therapist explores with the client why he has not attended therapy for an extended period of time. (d) T: When we had our last session was there something I said that freaked you out? C: Don’t remember T: I asked you how sex was with your boyfriend. Was there something we talked about that upset you or irritated you? 330 15. Frame Changes: Did the therapist address frame changes? It is important that any changes to therapy be addressed. There are many types of frame changes that can take place. Changes can be as significant as the therapist going on holidays resulting in a change in the pace of therapy or the client not attending a session or attending late. However, frame changes can be as small as the client exhibiting a new behaviour such as bringing food and drink into session or asking the therapist to open or close the window. With such changes, the therapist is expected to explore what these changes mean for the client. One of the main reasons for addressing frame changes is that clients with BPD frequently address difficult issues through actions. By addressing frame changes the therapist is addressing issues that may be taking place between the therapist and client. A frame change can also be a sign of a disjunction, such as the client failing to attend therapy because of something said by the therapist in the previous session. When the therapist attempts to repair such a disjunction, such an intervention needs to be rated under Disjunctions. If the therapist does not address a potential disjunction, the rater may rate under frame changes. It is also possible for the rater to rate under this item and ’disjunctions’ in the event that the therapist firstly checks with the client about a frame change and then mentions a possibility of a disjunction. EXAMPLE T: I tried to call you. I left a message when you did not come. C: I was sick. T: I rang your home number over a period of time. C: I was hiding in my room. I was not talking to anyone. T: So it was not about the therapy. C: No, I felt frozen. I could not talk to anyone. 331 16. Warmth: Did the therapist convey warmth? Warmth, which has been equated with unconditional positive regard, has been defined by Rogers and Truax (1967) as "the therapist communicating to his [sic] client a deep and genuine caring for him [sic]as a person with human potentialities, a caring uncontaminated by evaluations of his [sic] thoughts, feelings, or behaviors". This communication need not be explicit but the therapist's caring should be made evident by her/his behaviour. Raters who have developed their own operational definition of warmth are encouraged to use it only if it is consistent with how warmth is defined above. The rater must be careful not to assume that the therapist conveys warmth merely because she/he is a therapist. The rater must also remember that she/he is not rating how warm the therapist is in general, but rather how much warmth the therapist conveyed in the session being rated. The rater should begin at a default of 4 and rate upwards or downwards depending on how much warmth the therapist conveys to the client. Distinction The rater should rate Item 16 and Item 17 independently. It is possible for the therapist to be warm and caring and yet not get along with the client. Conversely, it is possible for the therapist to not demonstrate warmth or caring for the client and yet develop strong rapport. 332 17. Rapport: How much rapport was there between therapist and client (i.e., how well did the therapist and client get along)? This item is intended to measure the extent to which the relationship between the therapist and client is marked by harmony and accord (i.e., how well the therapist and client got along in the session). Raters who have developed their own operational definition of rapport are encouraged to use it only if it is consistent with how Rapport is defined above. Among the items in this scale, this item is clearly the most dependent on client behaviour as well as therapist behaviour. Although the rater should assign a low rating to this item if she/he believes that the therapist made efforts to get along with the client without success, this item should not be given a high rating unless rapport clearly existed between the therapist and client. 333 18. Agenda Setting: To what extent to the therapist set out an agenda for the session? Agenda setting involves the therapist negotiating with the client on how the session will be organised e.g. negotiating what time will be spent on various topics or interventions. While agenda setting can be a collaborative exercise, it is usually initiated by the therapist. Agenda setting is proscribed in the Conversational Model. Rather, the model aims to encourage the client to find a space in therapy where they can experience and reflect on their life events without being influenced by an agenda. However, it is appropriate for the therapist practising the CM to undertake agenda setting under three scenarios and therefore should not be rated. Firstly, in the event of a risk of deliberate harm to self or others is raised before or in session it is incumbent on the therapist to address these issues by the end of the session. Secondly, in a situation where there has been a major frame change, such as non-attendance in therapy over a period of time, the therapist will need to devote a part of the session to explore what this means for the client. Thirdly, if the therapist perceives that there is a serious threat to therapy such a disjunction, it is expected that the therapist will address such threats. EXAMPLES (a) T: It seems like it would be important for us to spend some time talking about your family. What do you think about doing some of that today? (b) T: What would you like to cover today in session? Distinction If the therapist sets out an agenda without any negotiation with the client, then the therapist’s behaviour should be rated under item 19 (Directiveness). 334 19. Directiveness: How much did the therapist direct or guide the session in an explicit way? The purpose of this item is to rate occasions when the therapist explicitly directed the session. The therapist might accomplish this by initiating a significant change in content or shift the focus of the session or by maintaining the focus on topics which she/he wants to discuss. However, it is appropriate in the Conversational Model for the therapist to be directive under three scenarios and therefore should not be rated under this item. Firstly, in the event that a risk of deliberate harm to self or others is raised before or in session, it is incumbent on the therapist to address these issues by the end of the session. Secondly, in a situation where there has been a major frame change such as non-attendance in therapy over a period of time, the therapist will need to devote a part of the session to explore what this means for the client. Thirdly, if the therapist perceives that there is a serious threat to therapy such as a serious disjunction, it is expected that the therapist will address such threats. In the event of any of the above three scenarios, a therapist should not be allocated any points in relation to interventions in the above three scenarios. EXAMPLES The following are examples of explicit guidance by the therapist: (a) T: All right. Let’s shift gears now and talk about what has been happening in the past week. (b) T: (After discussing how the client was getting along at home and on the job): What symptoms have you experienced since I last saw you? C: In general I have been irritable with my husband and the kids this week. T: It sounds as if we should discuss what has been going on at home but before we do that, how did the job interview go? (c) C: (In the midst of a discussion about a situation at the client's workplace that involved the foreman): Even my buddy at work, Jim, says that the foreman is hard to get along with. But that might be because Jim is missing a lot of work lately. He always wants me to cover for him. I am getting sick of being used by Jim. T: It sounds like you have some issues about Jim that you might want to talk about. However, before we do that, can we discuss your interactions with the foreman. 335 DISTINCTIONS Scoring under this item requires identifying when the therapist clearly changes the direction in the session. Generally, asking questions should be rated under item 24 involving gathering information. However, asking questions when it also involves a shift in the direction of therapy are rated under this item as well. It is also possible that some interventions can be rated under agenda setting (item 18) and under this item when the agenda setting appears to be an attempt to change the direction of the session. 336 20. Providing reassurance: Did the therapist provide reassurance to the client? This item is not part of the Conversational Model. A high score on this item would suggest that the therapist was not practising the CM. The reason for not using reassurance in the model is that it tends to move the client away from the experiences in session or experiences being described in session. Providing reassurance can close down exploration of feelings. Examples: The following would be rated high and therefore poorly against CM adherence. (a) C: My parents were furious about me dropping out of University. T: It was wrong of your parents to get mad with you. b) C: I am really a bad person. T: No, you’re not. (c) C: I’m really in trouble. T: No, you’re not actually. 337 21. Advice giving: Did the therapist provide any advice (including non- psychological and psychological advice) and undertake problem solving? This item rates the extent to which the therapist provided advice such as recommending the client undertake certain tasks or involve in problem solving. However, therapist interventions relating to suicidal or self-harming behaviour are not rated under this item. 338 22. Providing psychological techniques: Did the therapist offer psychological techniques to assist the client or suggest various types of practice of techniques between session? Examples of techniques are use of relaxation exercises, sensitisation and exposure interventions, pleasure scheduling, and thought monitoring. 339 23. Psychoeducation: Did the therapist provide psychoeducation around various issues affecting the client? Psychoeducation involves the therapist specifically referring to research or evidence about a psychological phenomenon. For example, the therapist could provide psychoeducation around giving up smoking and say to the client that “research has shown that quitting cigarettes is the hardest of all drugs to give up.” This item involves the therapist providing information about how people in general feel, behave, or respond regardless of whether the information applies specifically to the client. 340 24. Information gathering: Did the therapist gather information by way of questioning which was not for the purposes of clarification. Regular questioning of the client is not part of the Conversational Model. However, it is appropriate to ask the occasional question especially if it is to seek clarification from the client (such as a question relating to a gap in the client’s story) or relates to emotional aspects of the client’s story (such asking the client how he or she is affected by something that the client is describing such as an illness). It is important that appropriate questions relate to the content of what the client is saying especially the emotional content. Any other type of questions about the client’s life should be construed as information gathering. Distinctions Asking questions around the use of a psychological technique would be rated under providing psychological techniques. For example, asking questions about the content of a diary card or a homework task would be rated under delivery of a psychological technique or homework assignment respectively. Examples The following is an example of information gathering. C: I had a big exam today. T: What was the exam? C: It was an English exam. T: How do you think you went? The following are two examples of an appropriate question that would not be construed as information gathering followed by an information gathering example. C: I have been sick with asthma off and on for a long time. T: How does the asthma affect you? An information gathering respond could have been: T: What medication do you take? 341 25. Homework Assigned or reviewed: Did the therapist or client develop one or more specific assignments for the client to engage in between sessions? Did the therapist review any homework assigned to the client? The purpose of this item is to measure the extent to which the therapist develops homework or assists the client in developing homework. Homework can be an assignment which the client is to engage in (but not necessarily complete) before the next session. This item also measures the therapist’s efforts to review any homework negotiated with the client. References Sharpiro, D.A. & Startup, M.J. (1990). Rater’s manual for the Sheffield Psychotherapy Rating Scale. Memo 1154, MRC/ESRC, Social & Applied Psychology Unit, Department of Psychology, The University of Sheffield S102RN. Sharpiro, D.A., Barkham, M., Hardy, G.E., & Morrison, L.A. (1990). The second Sheffield psychotherapy project: Rationale, design and preliminary outcome data. British Journal of Medical Psychology, 63, 97-108. 342 Appendix A: Sheffield Psychotherapy Rating Scale Table 1: Items (17) which have been included in or adapted to the NASCoM (The items numbers are references to the SPRS item numbers) 1. Setting and following agenda: Did the therapist work collaboratively with the client to formulate and follow a specific agenda for the session? (Adapted) 2. Homework reviewed: Did the therapist review previously assigned homework with the client? (Adapted) 11. Warmth: Did the therapist convey warmth? 12. Rapport: How much rapport was there between therapist and client (i.e., how well did the therapist and client get along)? 13. Empathy: Was the therapist empathic toward the client (i.e. did she/he convey an intimate understanding of and sensitivity to the client's experiences and feelings)? (adapted) 18. Language of mutuality: Did the therapist use the language of shared endeavour ('I' and 'we')? 29. Patterns in Relationships: Did the therapist draw parallels or point out patterns in two or more of the client's relationships for the purpose of helping the client understand how she/he functions in interpersonal relationships? 30. Cue Basis: Did the therapist explicitly base his/her interventions on cues (verbal and non-verbal) supplied by the client? (Adapted) 31. Metaphor: Did the therapist encourage and elaborate the client's use of metaphor? 32. Focusing: Did the therapist focus on the here and now experience of the client in the session, encouraging the client to stay with feelings before any attempt to 'explain' them? (Adapted) 34. Disclosure: Did the therapist make appropriate use of self-disclosure to advance a shared understanding of events and processes in the therapy? (Adapted) 35. Limitations: Did the therapist promote the client's exploration of feelings concerning the limits to therapy, and boundary, loss and internalisation issues related to termination? 38. Explanatory Hypotheses: Did the therapist introduce possible reasons (which may or may not be 'causes') for the client's behaviour and experiences, particularly in respect of disturbances in relationships within and outside therapy? 343 51. Exploration Of Feelings: Did the therapist help the client to explore her/his feelings related to an interpersonal relationship? (Adapted) 52. Acknowledgment Of Affect: Did the therapist attempt to help the client to acknowledge affect that she/he was not expressing or of which she/he was unaware? (Adapted) 53. Acceptance Of Affect: Did the therapist encourage the client to accept feelings of which the client is aware but which are painful or uncomfortable? 72. Homework Assigned (Adapted) 344 Table 2: Items (41) in the SPRS not included in the NASCoM 7. Supportive Encouragement 8. Convey Expertise 9. Therapist's Communication Style 10. Involvement 14. Formality 16. Encourages Independence 17 Negotiating Style: Did the therapist express his/her views concerning the patient's experiences and circumstances as tentative statements, open to correction, and inviting elaboration and feedback? 21. Specific Examples 26. Exploratory Therapy Rationale 27. Relating Interpersonal Change To Therapy 33. Confrontation: Did the therapist confront the client with his/her avoidances by describing them and either directing the client to cease avoiding or facilitating the client's ceasing to avoid? 36. Understanding Hypotheses: Did the therapist offer statements of empathic understanding that brought her/his own perspective to bear in the mutual understanding of the client's experience? 37. Linking Hypotheses: Did the therapist link the client's present feelings with feelings in other contexts and at other times, with the central link being between each of these and the 'here and now' of the therapeutic relationship? 39. Sequencing of interventions 40. Structuring the session 54. Relationship of thoughts and feelings 55. Rationale for cognitive procedures 56. Relate improvement to cognitive change 57. Reporting cognitions 58. Exploring personal meaning 59. Recognizing cognitive excluded 60. Exploring underlying assumptions 61. Distancing of beliefs 62. Examine available evidence 63. Testing beliefs prospectively 64. Searching for alternative explanations 65. Realistic consequences 66. Adaptive/functional value of beliefs 67. Maintaining gains 68. Rationale for behavioural procedures 69. Practicing "rational responses" 70. Planning/practicing alternative behaviors 71. Skills training: did the therapist attempt to teach the client skills (e.g. Assertiveness, social skills, task relevant skills) in the session? 345 73. Increasing pleasure and mastery excluded 74. Scheduling/structuring activities excluded 75. Self-monitoring excluded 76. Recording thoughts excluded 77. Manipulating behavior via cues or consequences 78. Negotiating therapy content 79. Explanation for therapist's direction 80. Summarizing 346 Appendix A: Newcastle Adherence Scale for Conversational Model (NASCoM) 1. Receptive listening: Did the therapist appear to allow silence to continue (or use minimal encouragement such as "okay or ""uh-huh, ") as a means of encouraging the client to talk 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 2. Tentative style: Did the therapist express his/her views concerning the patient's experiences and circumstances as tentative statements, open to correction, and inviting elaboration and feedback? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 3. Language of mutuality Did the therapist use the language of shared endeavour ('I' and 'we' and passive tense)? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 4. Identifying Patterns in relationships: Did the therapist draw parallels or point out patterns in two or more of the client's relationships for the purpose of helping the client understand how she/he functions in interpersonal relationships? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 5. Focus on Client-Therapeutic relationship: Did the therapist address the client's present feelings around the therapeutic relationship or make links with the client’s present feelings about the therapeutic relationship with feelings in other contexts and at other times? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 6. Emotional Attunement: Did the therapist attune to the emotional cues or other words offered by the client by using either of the following micro-skills: coupling, and amplification? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 347 7. Awareness of Feelings: Did the therapist encourage client to consider unspoken feelings of which the client may be unaware or avoiding. 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 8. Avoidance of Affect: Did the therapist assist the client address any avoidances experienced by the client? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 9. Acceptance of affect: Did the therapist encourage the client to accept feelings of which the client is aware but which are painful or uncomfortable? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 10. Explanatory Statements: Did the therapist introduce explanatory statements which offer possible reasons for the client's behaviour and experiences 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 11. Metaphor: Did the therapist encourage and elaborate the client's use of metaphor? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 12. Personal disclosure: Did the therapist disclose or respond to client’s personal questions about the therapist in way that advanced a shared understanding of events and processes in the therapy? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 13. Limitations: Did the therapist promote the client's exploration of feelings concerning the limits to therapy, and boundary, loss and internalisation issues related to termination? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 348 14. Disjunctions: Did the therapist address any disjunctions in therapy? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 15. Frame Changes: Did the therapist address frame changes? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 16. Warmth: Did the therapist convey warmth? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 17. Rapport: How much rapport was there between therapist and client (i.e., how well did the therapist and client get along)? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 18. Agenda Setting: To what extent to the therapist set out an agenda for the session? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 19. Directiveness: How much did the therapist direct or guide the session in an explicit way? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 20. Providing reassurance: Did the therapist provide reassurance to the client? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 21. Advice giving: Did the therapist provide non-psychological advice and undertake problem solving? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 349 22. Providing psychological techniques: Did the therapist offer psychological techniques to assist the client or suggest various types of practice of techniques between sessions? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 23. Psychoeducation: Did the therapist provide psychoeducation around various issues affecting the client? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 24. Information gathering: Did the therapist gather information by way of questioning which was not for the purposes of clarification. 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 25. Homework assigned: Did the therapist or client develop one or more specific assignments for the client to engage in between sessions? 0 1 2 3 4 5 6 not at all 0-5% 6-10% 10-20% 21-25% 26-30% 30%- 350