Purpose: Therapists are often charged with the seemingly impossible task of predicting their client’s future behavior, particularly behavior that may result in harm or death. Adverse events (AE) refer to a constellation of behaviors or events that interfere with treatment and exhibit a risk to the safety of the patient, which include suicide attempts, non-suicidal self injury (NSSI) and suicidal ideation. This is the first study that seeks to identify and associate in-session markers in DBT prior to AEs.
Method: The proposed study sought to identify whether ruptures in therapeutic alliance (3RS; Eubanks-Carter, Muran & Safran, 2015), the frequency and intensity of negative-self referential speech (LIWC2015; Pennebaker, Booth, Boyd & Francis, 2015) and periods of psychomotor agitation are associated with AEs within a course of Dialectical Behavior Therapy (DBT). By coding videotaped psychotherapy sessions (n = 98) across 21 patients diagnosed with Borderline Personality Disorder (BPD), the researchers prospectively examined the association between in-session phenomena during the session prior to an AE. Exploratory logistic multilevel modeling, mean comparison and latent profile analysis (LPA) techniques were used to identified in-session markers associated with adverse events across the course of DBT treatment.
Results: Using a multilevel model building approach to account for the nested structure, increases in content/affect split was associated with increased likelihood (36% increase in log-odds) of NSSI occurrence reported in the subsequent session when controlling for frequency of past NSSI episodes. When controlling for prior suicide attempts, withdrawal and confrontation ruptures did not predict the occurrence of suicide attempts in the subsequent session. To further examine the heterogeneity of the Level 1 variables (i.e., in-session markers), the LPA fitted afive-profile solution that captured relative differences in mean frequencies of coded markers.The latent “session types” were named based on their in-session characteristics, with AEs identified post-hoc within the identified profiles. While AEs were distributed across multiple profiles, visual inspection aligned with the findings in the multilevel model. Sessions characterized by elevations in content/affect split and behaviors that distance from the therapist preceded NSSI during treatment. The majority of the sessions prior to suicide attempts (70%) during the study period were assigned to the profile with the lowest mean frequency of in-session markers.
Clinical implications: The strength of the therapeutic alliance in DBT is an essential component of effective treatment. Therapeutic ruptures, particularly withdrawal ruptures, occur frequently in DBT treatment. Attending to these ruptures, especially occasions when a patient’s affect and verbal content are not congruent, may signal to the therapist that the patient requires additional support. In-session content/affect split may represent a vulnerability factor that puts the patient at increased risk of NSSI behavior due difficulty attuning to their internal experiences and limitations in their emotional flexibility.
Limitations: Similar to other studies that examine self-harm, the low base-rate of suicide attempts and NSSI behavior complicates empirical study. Since the study utilized strict inclusion criteria for only individuals diagnosed with BPD, findings cannot be generalized to patients with other psychiatric diagnoses. While some therapist effects are controlled for in the study since one therapist treated all the patient included in the study, the study does not account for therapist factors that may influence the therapy dyad. Given the limited sample size, there was not adequate power to fit more complicated models (e.g., inter-level and intra-level interactions, random effect predictor variables, etc.).
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/d8-t1e0-xa21 |
| Date | January 2019 |
| Creators | Lynch, David Alexis |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0024 seconds