Examining Therapeutic Alliance as a Mechanism of Change in Dialectical Behavior Therapy for Subgroups of Individuals With Borderline Personality Disorder

Millions of people worldwide are diagnosed with borderline personality disorder (BPD), a complex mental disorder characterized by significant affect lability, an unstable sense of identity, tumultuous interpersonal relationships, and difficulties managing impulsive behaviours. According to the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders, to be diagnosed with BPD, individuals need to present with a minimum of 5 of 9 possible psychiatric symptoms, resulting in 256 possible combinations of symptoms. Given the significant symptom heterogeneity, recent research has examined whether reliable, homogenous subgroups of BPD symptoms exist. If so, it may be that individuals within each subgroup would benefit from tailored modifications to current interventions (namely, dialectical behavior therapy, or DBT) for the treatment of their unique subset of symptoms. In a recent study in our laboratory at the BPD Clinic, Antoine and colleagues (2022) applied a latent class analysis to existing data on the symptoms of individuals diagnosed with BPD and identified three BPD symptom subgroups, or "classes" of BPD symptoms, which are termed the interpersonally unstable, nonaffective labile, and dissociative/paranoid classes, respectively. In the current dissertation, these classes are examined as predictors of psychiatric functioning after individuals with BPD completed 12 months of standard DBT in outpatient mental health hospitals. Moreover, therapeutic alliance, which is defined as the strength of the relationship between the therapist and their patient, is examined as a potential mediator of the relationship between class membership and psychiatric functioning posttreatment. Despite hypotheses that class membership would be predictive of treatment outcome after 12 months of DBT, there were no significant relationships between class membership and psychiatric functioning, deliberate self-harm, or days in treatment, and mediation with the therapeutic alliance did not occur. Nonetheless, DBT was effective across the symptom classes and therapeutic alliance predicted better outcomes. Specific reasons as to why I failed to find meaningful relations between symptom classes and outcomes, clinical implications, and future directions are explored. This dissertation is the first to examine how symptom class inclusion relates to DBT treatment outcomes as mediated by therapeutic alliance.