Evaluating therapeutic alliance longitudinally: describing therapeutic alliance growth and its implications for outcomes

Abstract

Therapeutic alliance, the helping relationship that develops between client and therapist, is the most commonly studied process variable in mental health treatment. In child/adolescent treatment, the relationship that develops between therapist and the youth’s caregiver has also been argued as important. This is due to the routine involvement of caregivers in treatment tasks (e.g., learning new parenting skills), as well as their involvement in dictating the youth’s treatment goals and treatment agenda. Alliance is theorized not to remain static; it deteriorates or grows stronger over the course of treatment. Yet despite alliance’s theorized volatility and instrumental role in predicting eventual treatment response, very little research has investigated whether longitudinal alliance change within client relates to outcome. Using data from the Multimodal Treatment Study of Children with AD/HD, caregivers participating in the parent-training arm of the study were utilized to investigate how alliance growth affects outcome using statistical methods that can model appropriately within-client change (hierarchical linear modeling, survival analysis). Consistent with hypotheses, caregiver-reported alliance was found to be significantly volatile within-clients and has different trajectories in early versus late treatment. These findings call into question the utility of the alliance literature’s traditional means of summarizing therapeutic alliance (e.g., average overall alliance, one-session snapshots of alliance over phases of treatment). Contrary to the other hypotheses, treatment engagement variables did not consistently predict eventual treatment response, patterns of missing alliance data, or early termination from the study. The only consistent predictor of outcome (i.e., treatment response and early termination), was treatment group assignment, which is consistent with the main findings of the study. Explanations for the results of this study are generated, and the conclusion of the study outlines the new directions that can move the therapeutic alliance literature forward using advances in longitudinal statistics.