Alliance Ruptures and Psychotherapy Outcomes: A Multilevel Meta-Analysis of Their Association

Although two meta-analyses have examined the relationship between rupture repair and treatment outcome, no meta-analysis to date has investigated the association between alliance ruptures and treatment outcome. Numerous primary studies have explored this relationship; however, the methods used to assess ruptures vary considerably. These include direct measures, such as ratings of the occurrence, frequency, or intensity of ruptures by observers, patients, or therapists, and indirect measures, such as fluctuations in alliance ratings. This methodological heterogeneity may contribute to inconsistent findings across studies. The present meta-analysis sought to clarify the association between alliance ruptures and treatment outcomes by including only studies that assessed ruptures using direct measures.

After applying inclusion and exclusion criteria, four publications remained reporting 18 effect sizes of the associations between ruptures and outcomes. Those provided data from N = 301 clients undergoing psychotherapy. Multilevel models with a three-level structure with restricted maximum likelihood to estimate the aggregated effect size of ruptures on outcome were conducted.

The results showed a significant aggregated effect size of z = -0.22, SE = 0.04, 95 CI [-0.303, -0.133]; t(17) = -5.40, p < 0.001, r = -0.21, indicating that alliance ruptures were associated with poorer therapy outcomes.

A small but consistent negative association between alliance ruptures and treatment outcomes was found. An individual patient data meta-analysis would be an important future endeavor in the field of alliance ruptures, by including more studies and a homogenous analytic approach to estimate rupture-outcome associations and including rupture-repair as a moderator.