Barber et al., (2000) Alliance predicts patient’ outcome beyond in-treatment change in symptoms. Journal of Consulting and Clinical Psycholgoy, 68,1027-1032.

This article evaluated whether the strength of the therapeutic alliance was predictive (“causally related”) of outcome independent of improvements in depression (BDI) from the beginning of therapy until the alliance was measured.

88 clients in separate clinical trials for treatment (supportive/expressive dynamic therapy) of either (a) chronic depression, (b) generalized anxiety disorder, or (c) either avoidant or obsessive compulsive personality disorder answered the BDI at each individual session and completed a measure of the strength of alliance (the CALPAS) at weeks 2, 5, and 10 of therapy.

Important adjusted scores used in the analyses were the SUBSEQUENT CHANGE SCORE, which was a residualized termination BDI score adjusted for the BDI score in which alliance was assessed; and the PRIOR CHANGE SCORE, which was the residualized session BDI score adjusted for the intake BDI score.

Mean BDIs improved over sessions: Intake=20.4, week 2=16.0,

Week 5=14.3, week 10=12.5, month 4=10.8, and termination=9.3.

CALPAS predicted residualized SUBSEQUENT CHANGE in depression at 4 mo. and at termination, after partialing out PRIOR CHANGE in depression. Correlation ranging from r = .21 to r = .33, all sig.

PRIOR CHANGE predicted CALPAS scores at sessions 5 and 10, but not session 2, probably because there was insufficient improvement in depression by Session 2 to affect ratings of the strength of the alliance.