Aurelie Lange

98 | Chapter 6 Abstract Objective: The client-therapist working alliance and therapist adherence to the treatment manual are both associated with treatment outcomes. We investigated how alliance and adherence develop during treatment and how this development is uniquely and jointly related to treatment outcomes up to 18-months post-treatment using a variable-centered (latent growth curve modeling) and a person-centered approach (latent class growth analysis). Method: We used routinely collected data from 848 adolescents (66% male and 76% western, mean age = 15.25 years) and their caregivers participating in Multisystemic Therapy (MST), a family- and community-based intervention for antisocial adolescents. Alliance and adherence were measured monthly through phone interviews with the caregivers using the Therapist Adherence Measure-Revised. Outcomes were assessed at the end of the treatment and at 18-months post-treatment using the scale ‘rule-breaking behavior’ of the Child Behavior Checklist and two MST Ultimate Outcomes (i.e., police contact and out-of- home placement). Results: On average, alliance and adherence showed an increasing and then flattening slope. We identified two trajectory classes for alliance and three classes for adherence, which were mainly characterized by different initial levels of alliance and adherence, respectively. Both alliance and adherence predicted treatment outcomes at the end of treatment, but not at 18 months post-treatment. The effects of alliance could not be replicated using the person-centered approach. Conclusions: We advocate the need to consider the dynamic nature of adherence in research as well as clinical practice. The lack of a robust alliance-outcome relationship may be a consequence of the small variance observed in alliance.