Public significance statement: The present study shows that within a transdiagnostic sample of 430 patients treated by 151 therapists, the alliance-outcome relation is dependent on the number of aggregated alliance assessments as well as on the type of outcome assessment (i.e. distress vs. subjective change measures). Aggregated early alliance assessments were shown to relate most strongly to retrospectively evaluated subjective change at post-treatment. Examining associations with subjective change measures provide additional clinical perspectives on process and outcomes when conducting clinical trials and routine outcome monitoring.

Public abstract: Despite ongoing encouragement to systematically consider patient-relevant outcomes in clinical trials of psychotherapy, subjective change measures (e.g., how patients evaluate the change during psychotherapy) are still neglected in guidelines of how to conduct trial research. The present analysis shows that these measures are able to detect strong process-outcome associations in comparison to distress measures and have the potential to substantially deepen our psychological knowledge of how to understand successful psychotherapies but also successful psychological treatments more generally.

Authors note: The present data set refers to Flückiger et al., 2013; preliminary results of the present analyses were presented at the Conference of the International Society of Psychotherapy Research (SPR) at June 2018 at Amsterdam.

Abstract

Objective: Despite meta-analytic evidence showing that alliance is associated with post-treatment outcomes, several open questions still remain regarding this relation and its measurement. The current study sought to address three such questions simultaneously by investigating the alliance-outcome relation with two distress measures and four subjective change measures at post-treatment separately: First, we investigate whether (or not) the progressive aggregation of the early alliance assessments will increase the overall alliance-outcome relation across the six outcome measures. Second, we investigate whether the alliance-outcome relations using subjective change measures are independent from intake distress and early response. Third, we explore whether the progressive aggregation of the alliance on outcomes becomes particularly apparent between therapists or within therapists again investigating distress as well as subjective change measures.

Method: Data were drawn from n = 430 patients treated by n = 151 therapists. Patient ratings of early alliance were assessed after session 1 to 6. Six outcome measures were considered. Patients completed two commonly used distress measures at intake and at post-treatment and four measures of retrospectively evaluated subjective change at post-treatment. Multilevel models were used to predict treatment outcomes from alliance ratings aggregated across multiple sessions.

Results: The proportion of variance in outcome predicted by alliance scores varied considerably depending on the number of alliance assessments which were aggregated, as well as on the type of outcome assessment (distress vs. subjective change measures) explaining up to 15% of outcome variance. Improvements in the strength of prediction with aggregated alliance assessments were most pronounced for subjective change measures for between-therapist components of the alliance.

Conclusion: The strength of the alliance-outcome relation varies depending on the number of alliance observations that are aggregated, on how outcome is assessed (distress vs. subjective change measures), and whether scores are investigated at the between- or within-therapist level. Examining associations with subjective change measures provides an additional, person-centered metric of the relation between early alliance and treatment outcomes.