Testing Borderline Personality Disorder Diagnosis as a Moderator of the Longitudinal Course of Alliance and Outcome in a Community Mental Health Center
Open Access
- Author:
- Clouthier, Tracy Lynn
- Graduate Program:
- Psychology
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- December 12, 2018
- Committee Members:
- Kenneth Levy, Dissertation Advisor/Co-Advisor
Kenneth Levy, Committee Chair/Co-Chair
Louis Georges Castonguay, Committee Member
Jeffrey Hayes, Committee Member
Sy-Miin Chow, Outside Member - Keywords:
- psychotherapy
borderline personality disorder
therapeutic alliance - Abstract:
- The therapeutic alliance is the most commonly studied psychotherapy process variable (Castonguay et al., 2006) and has been consistently associated with outcome (e.g., Horvath et al., 2011). However, there is some debate as to the direction of this relationship (e.g., Barber et al., 2000; DeRubeis et al., 2005), and recent research suggests that alliance and outcome may influence one another in a reciprocal manner throughout treatment (Falkenström et al., 2014; Xu & Tracey, 2015). Further, the relationship between alliance and outcome may differ based on diagnosis (Flückiger et al., 2013). The interpersonal dysfunction and affective instability characteristic of borderline personality disorder (BPD) may have a particularly strong impact on the alliance (Hirsh et al., 2012; Levy et al., 2010), and results of a meta-analysis suggest that the alliance-outcome relationship is somewhat weaker in this population (Levy et al., in review). The primary goal of the present study was twofold. First, we attempted to replicate recent studies suggesting that the relationship between alliance and outcome is a reciprocal one. Second, we aimed to examine how the alliance changes over time in psychotherapy for BPD and how it relates to symptom change. We also explored criterion-based and control chart methods for detecting alliance ruptures using patient-reported alliance scores and how they performed for patients with BPD and other patients in the present sample. A sample of 40 outpatients was recruited from the Penn State Psychological Clinic. Participating patients completed the Brief Alliance Inventory (BAI; Mallinckrodt & Tekie, 2016) after each session for up to 30 sessions, in addition to completing the Treatment Outcome Package (TOP; Kraus, Seligman, & Jordan, 2005) prior to each session, as is standard practice at the clinic. These aims were examined using a bivariate (alliance and symptoms) multilevel model with cross-lagged and autoregressive associations, as well as between-group comparisons of means and variance, and criterion-based and control chart methods for detecting alliance ruptures. Contrary to our hypothesis that the relationship between alliance and symptoms would be reciprocal, lower symptoms at one session were predictive of higher alliance at the following session, but alliance at one session did not predict symptoms at the following session. Also contrary to our hypotheses, BPD diagnostic status was not found to impact the cross-lagged associations between alliance and symptoms or the rate of alliance ruptures detected using criterion- and control-chart-based methods, though patients with BPD did exhibit more variability in alliance scores than did other patients. The implications of these findings for research and practice, the strengths and limitations of the study, and directions for future research are discussed.