Change in therapy: A model of the effects of systematic treatment selection variables and quality of alliance on individual growth

A rift exists in psychotherapy research and literature between the roles attributed to common and specific factors in accounting for positive change during treatment. This rift has created a challenge for clinicians that desire to engage in a research-informed practice. The purpose of this study was to address the disparity between research findings on the contribution of the alliance to therapy outcome, and the evidence of the benefits of Systematic Treatment Selection (STS; Beutler, Clarkin and Bongar, 2000). It tested a Latent Growth Model that described individual change in depression and substance use in 190 depressed, substance-abusing patients from two Randomized Clinical Trials. In addition, the model tested the impact of the alliance and STS variables on that change, and advanced the premise that initial treatment matching to patient characteristics beyond diagnosis improves the quality of the alliance. Patients (48% females; 19% ethnic minorities) had been randomly assigned to one of five 20-session-long treatments, which included couples cognitive-behavioral, family-systems, cognitive, narrative, and prescriptive therapy. While dropout from treatment was moderate (48.4%), a good fit was found for a quadratic function of change for depression (chi2 = 16.47, df = 9, p = 0.059; CMIN/DF = 1.83; NNFI = 0.986; CFI = 0.991; RMSEA = 0.06), with significant symptom reduction during treatment and small increase at follow up. A quadratic function was also found to best fit the data for substance use (chi2 = 16.98, df = 12, p = 0.15; CMIN/DF = 1.42; NNFI = 0.996; CFI = 0.997; RMSEA = 0.04). While regression coefficients of some of the added predictors were not found to reach statistical significance, the complete predictive models yielded acceptable fit indexes for both depression (chi2 = 112.53, df = 33, p < 0.05; CMIN/DF = 3.41; NNFI = 0.942; CFI = 0.971; RMSEA = 0.10) and substance use (chi2 = 94.77, df = 37, p < 0.05; CMIN/DF = 2.56; NNFI = 0.970; CFI = 0.983; RMSEA = 0.08). Sustained differences in outcome among treatment modalities did not reach significance. However, the match between patient subjective distress at treatment entry with interventions that addressed affective experiences predicted a better outcome for substance use. Furthermore, initial match of patient coping style (internalizing vs. internalizing) with treatment focus (insight vs. symptoms) predicted a better alliance for both outcome variables. Thus, evidence was found in support of the conclusion that common and specific factors are far from independent.