Relationship processes in youth psychotherapy: measuring alliance, alliance-building behaviors, and client involvement.(Report)

This study explores hypothesized associations among therapist engagement strategies, therapeutic alliance, client involvement, and treatment outcome in a randomized clinical trial comparing cognitive behavioral psychotherapy and nondirective supportive psychotherapy for adolescents with depressive symptoms who have attempted suicide. Ratings from audiotapes and self-report of the first four sessions for 23 adolescent clients were used. It was expected that therapeutic relationship variables would be equally important in both treatments. However, preliminary evidence appeared to be emerging only for therapeutic alliance and client involvement being related to treatment outcome in the cognitive behavioral treatment. Therapist lapse behaviors were found to predict alliance across both treatments. On the other hand, there was some preliminary evidence for different therapist behaviors to be related to the therapeutic alliance in each treatment. Results suggest that there may be variation in effective relationship factors, depending on the specific therapeutic approach.

Keywords: therapeutic alliance; psychotherapeutic processes; involvement; therapist behaviors; adolescent psychotherapy

**********

The importance of therapeutic relationship variables has long been emphasized by practicing clinicians, especially among those who treat adolescents (Bickman et al., 2000; Meeks & Bernet, 2001). In support of this clinical perspective, a meta-analysis by Shirk and Karver (2003) revealed a small but reliable association between therapeutic relationship variables and treatment outcomes in child and adolescent therapy. The average effect size (r = .27) was almost identical to effect sizes found in two meta-analyses of adult psychotherapy (Horvath & Symonds, 1991; Martin, Graske, & Davis, 2000). However, despite the theoretical and clinical importance of the alliance in the adolescent treatment literature, only a handful of studies have evaluated alliance and other related relationship constructs in individual outpatient adolescent treatment (Hagborg, 1991; Hogue, Duaber, Stambaugh, Cecero, & Liddle, 2006; McNally & Drummond, 1973; Taylor, Adelman, & Kaser-Boyd, 1986; Tetzlaff et al., 2005). In addition, Shirk and Karver's (2003) review revealed substantial methodological problems in the empirical literature and little consensus regarding the measurement of relationship processes in youth psychotherapy. Hogue et al. (2006) were one of the first to examine alliance-outcome relations in separate treatments with adolescents participating in a randomized trial; however, they did not find early adolescent alliance to predict positive treatment outcomes in cognitive behavioral therapy (CBT) or multidimensional family therapy. Karver, Handelsman, Fields, and Bickman (2005) indicated the need for studies that further our knowledge on how different therapeutic relationship variables contribute to treatment outcome.

In a recent study, Chu and Kendall (2004) found child involvement in treatment tasks to be a significant predictor of clinical treatment gains. Related, Karver, Handelsman, Fields, and Bickman (2006) found a moderate relationship between treatment participation and therapeutic outcome across 10 youth treatment studies. It has been suggested further that client involvement in therapeutic tasks is likely facilitated by a strong therapeutic alliance (Chu & Kendall, 2004; Shirk & Karver, 2006); however, the association between alliance and involvement has rarely been evaluated in the youth treatment literature (Colson et al., 1991; Shirk & Saiz, 1992). According to this perspective, a positive relationship between adolescent and therapist promotes involvement in therapeutic work (Shirk & Russell, 1996). By client involvement, we mean cooperating with, being involved in, making suggestions about, and/or completing therapeutic tasks (e.g., homework, discussing feelings, responding to therapist requests; Karver et al., 2005). Hill (2005) contends that we need better measures of client involvement and proposes that involvement is intertwined with the therapeutic alliance and therapist engagement interventions but conceptually and methodologically separable. Measurement of each construct is essential if we are to identify the relative contribution of these processes to treatment outcome.

By the therapeutic alliance, we refer to the relational, emotional, and cognitive connection between the youth client and a therapist (e.g., bond, trust, feeling allied, and positive working relationship; Karver et al., 2005). Alliance is distinguished from client involvement on specific treatment tasks. At this point, it is not clear whether the adolescent alliance is best assessed from the perspective of the client or an observer (Shirk & Karver, 2003), as few studies have included multiple perspectives on the alliance (e.g., Creed & Kendall, 2005). It is important to consider a dual approach to examining the adolescent's experience of the relationship through self-report and behavioral markers that can be observed in therapeutic interaction. Friedlander et al. (2006) suggest that observer ratings may actually complement client self-report ratings. It is also important that the alliance be assessed in a developmentally appropriate manner, as most therapeutic alliance measures used in youth research have been borrowed from the adult treatment literature (Martin, Romas, Medford, Leffert, & Hatcher, 2006; Shirk & Karver, in press). Thus, this study utilized a commonly used self-report alliance measure, the Working Alliance Inventory (WAI; Horvath & Greenberg, 1986), and a developmentally appropriate observational coding system of the therapeutic alliance (Karver, Shirk, Day, Fields, & Handelsman, 2003) to assess the alliance and its predictive relationship with client involvement.

Considering the importance of the therapeutic alliance and client treatment involvement, it would be beneficial to identify potential determinants of these critical therapy variables. At present, little is known about factors that contribute to a strong client-therapist alliance or that foster client participation in treatment (Connolly Gibbons et al., 2003; Shirk & Karver, 2006). Most research on alliance formation has focused on client pretreatment characteristics (e.g., Clarkin & Levy, 2004; Colson et al., 1991; Connolly Gibbons et al., 2003; Eltz, Shirk, & Sarlin, 1995; Fields, Handelsman, Karver, & Bickman, 2004; Gaston, Marmar, & Thompson, 1988). There has been much less research examining therapist characteristics and behaviors that predict the therapeutic alliance, and most studies have used retrospective rather than prospective designs (Ackerman & Hilsenroth, 2003; Carroll, 2001; Horvath & Bedi, 2002; Weinberger, 2002). One would expect that therapist behaviors or strategies affect the development of the alliance and client involvement early in treatment as the therapist attempts to engage the often reluctant adolescent into the therapy process (Frank, Frank, & Cousins, 1993; Hoyt, 1996; Karver et al., 2005). In the child and adolescent therapy literature, there are even fewer studies than there are in the adult literature addressing specific strategies for engaging individual youth clients in treatment. Research has focused on various pretreatment strategies to prepare parents and children for treatment (Bonner & Everett, 1982; Coleman & Kaplan, 1990; Day & Reznikoff, 1980; Holmes & Urie, 1975; Kourany, Garber, & Tornusciolo, 1990; Rotheram-Borus, Piacentini, Cantwell, Belin, & Song, 2000; Rotheram-Borus, Piacentini, Miller, et al., 1996; Rotheram-Borus, Piacentini, Van Rossem, et al., 1996; Shuman & Shapiro, 2002; Tinsley, Bowman, & Ray, 1988; Wenning & King, 1995). Most of these studies showed that prepared children and parents evinced greater knowledge about therapy than did their unprepared counterparts; however, evidence linking this increased pretreatment knowledge to increased engagement in treatment remains quite mixed and typically only focuses on treatment attendance and not on actual involvement in session (Nock & Ferriter, 2005; Shirk & Russell, 1996). More recently, there has been an increasing pool of studies, particularly in family-based treatment research, examining variations of procedures during treatment that might enhance engagement (e.g., Coatsworth, Santisteban, McBride, & Szapocznik, 2001; McKay, Stoewe, McCadam, & Gonzales, 1998; Nock & Kazdin, 2005; Prinz & Miller, 1994; Robbins et al., 2006). The results of these studies have been promising in that they suggest that modifications of treatment processes and techniques can have an impact on treatment attendance, but again, there has not been a focus on actual within-session treatment involvement.