The Role of Working Alliance in the Treatment of Borderline Personality Disorder

Open Access
Wasserman, Rachel H.
Graduate Program:
Doctor of Philosophy
Document Type:
Date of Defense:
July 20, 2010
Committee Members:
  • Kenneth Levy, Dissertation Advisor
  • Kenneth Levy, Committee Chair
  • Aaron Lee Pincus, Committee Member
  • Louis Georges Castonguay, Committee Member
  • Peter Cm Molenaar, Committee Member
  • borderline personality disorder
  • alliance
  • psychotherapy research
  • HLM
The primary aim of the proposed research is to understand a proposed mechanism of change in psychotherapy for Borderline Personality Disorder (BPD) through examination of the relationship between alliance and outcome in two well established treatments for BPD. BPD is costly, painful, debilitating, and deadly, and thus, represents a serious clinical and public health concern. Across multiple studies of Axis I psychopathology, alliance has been shown to be a robust predictor of outcome. In addition, formation of an alliance has been written about extensively in theoretical descriptions of the treatment of BPD. However, little systematic research has been conducted to determine the relationship between alliance and outcome in BPD. The maximization of efficacy in the treatment of BPD depends upon understanding the specific mechanisms that characterize the disorder and that determine clinical change. Here, we examined working alliance in BPD, with a well standardized observer-rated measure (WAI) and a well characterized and representative sample of reliably diagnosed patients. We used archival data from a randomized clinical trial, including well characterized and standardized treatments to minimize competence as a confound for alliance. Through direct investigation of the alliance in a well controlled study of treatments for BPD, we begin to answer important questions regarding the role of alliance in therapeutic change with this population. The long-term goal of this research program is to elucidate the technical, relational, and contextual mechanisms of change in the treatment of specific psychological disorders, and to facilitate the translation of these findings into the validation and dissemination of efficacious treatments. Three foundational questions regarding the alliance in BPD were posed: 1) do treatment differences exist in the formation of early alliance, in line with emphasis on supportive and validating techniques early in treatment? 2) is alliance predictive of treatment retention/termination in this population? And 3) is alliance predictive of treatment response in BPD? Analysis of variance, survival analysis and hierarchical linear models were used to address these questions. It was found that working alliance was positive and equivalent treatments indicating that despite technical differences in approach, that therapists in all three treatment conditions were able to foster generally positive working relationships with their patients in the initial sessions of therapy. With respect to early termination, treatment type was found to be marginally associated with risk of dropout while working alliance was found to be significantly predictive of treatment retention. Hierarchical linear models were used to investigate the relationship of working alliance to initial symptomatic distress and functional impairment as well as to change during treatment. Consistent with predictions, working alliance was found to be associated with functional impairment but not symptomatic distress at the start of treatment. Working alliance was predictive at trend levels of change in global severity of symptoms across treatments. In addition, with respect to global assessment of functioning a trend level three way interaction among working alliance, treatment condition and time was found indicating that working alliance is significantly more related to change in global functioning in TFP than in SPT and marginally more related to change in global functioning in DBT than SPT. Working alliance was not found to predict change in the four other outcome variables investigated. Implications of these findings, strengths and limitations, and future directions for research are discussed.