Citation

Gamache D, Savard C, Lemelin S, Côté A, Villeneuve E. Compr. Psychiatry 2017; 80: 14-23.

Affiliation

CERVO Research Center, 2601 Chemin de la Canardière, Quebec City, QC, Canada; Institut universitaire en santé mentale de Québec, Centre de traitement le Faubourg Saint-Jean, 2601 Chemin de la Canardière, Quebec City, QC, Canada.

Copyright

(Copyright © 2017, Elsevier Publishing)

DOI

10.1016/j.comppsych.2017.08.001

PMID

28915423

Abstract

OBJECTIVE: Psychological treatment for patients with personality disorders (PD) is plagued with a high proportion of early dropouts, and attempts to identify risk factors for attrition have generated very few conclusive results. The purpose of the present study is to identify significant predictors of early treatment termination in a long-term psychotherapy program for PD.

METHODS: Data was retrospectively retrieved from 174 files of patients who began long-term psychotherapy in an outpatient treatment program in Quebec City, Canada. Socio-demographic, initial disturbance, and diagnostic variables were considered for prediction, along with a measure specifically designed to identify PD patients at risk of dropping out early from psychotherapy, the Treatment Attrition-Retention Scale for Personality Disorders (TARS-PD). Survival analysis using Cox proportional hazard regression was performed to identify significant predictors.

RESULTS: Results using univariate Cox proportional hazards regressions revealed that unemployment, Global Assessment of Functioning scores, and recent hetero-aggressive behavior were significant predictors of early dropout in the first six months of therapy. Adjusting for these three confounders, four of the factor scores from the TARS-PD (Narcissism, Secondary gains, Low distress, and Cluster A features) were significantly associated with dropout in univariate Cox proportional hazards regressions. Secondary gains and Narcissism remained significant predictors after entering all five TARS-PD factors in a multivariate Cox proportional hazards regression adjusting for confounders.

CONCLUSIONS: Taking into consideration specific treatment prognosis variables, such as those measured by the TARS-PD, might be more useful for dropout prediction in PD patients in comparison with more general demographic and diagnostic variables.

Copyright © 2017 Elsevier Inc. All rights reserved.

Language: en