Harnessing the potential of the therapeutic alliance

Abstract

Originally developed within psychoanalytic psychotherapy, the therapeutic alliance has emerged as a widely studied pantheoretical change variable. In fact, a Google Scholar search for “therapeutic alliance” and “outcome” returns roughly 135,000 hits. The vast majority of these studies have reported on the relationship between the alliance and psychotherapy outcome. However, a smaller but growing literature has also examined the relationship between the therapeutic alliance and pharmacotherapy adherence and outcome. The therapeutic or working alliance involves an ongoing collaboration between patient and provider encompassing both task- and affectively-oriented features of their relationship. Specifically, it is often defined as: a) the extent of patient-provider agreement on treatment goals; b) collaboration on treatment tasks necessary for goal attainment; and c) the affective bond (e.g., caring, liking, trust) between patient and provider 1. Patient and provider-rated, as well as observer-rated instruments are represented in the alliance-outcome literature. Consistent with previous findings, a recent meta-analysis 2 involving 190 studies reported a significant relationship between the therapeutic alliance and psychotherapy outcome (weighted r=.28, p<0.0001). Overall, meta-analytic findings reveal that the magnitude of the alliance-outcome relationship is modest, accounting for 5-8% of the variance in outcome. However, the relationship is consistent, having been demonstrated across a variety of populations and treatments, including individual, couples/family as well as child and adolescent psychotherapy. Although the literature linking the therapeutic alliance and psychotherapy outcome is robust, it is characterized by methodological shortcomings that have raised questions about the causal relationship between the alliance and symptom outcomes. Extant findings are observational rather than experimental. Inferring a causal relationship between alliance and outcome in such studies requires temporal precedence 3. That is, the alliance must predict subsequent outcome, while accounting for change that occurred prior to the alliance assessment 3. Many studies in this literature have not ruled out the possibilities that the alliance is “predicting” change that has already occurred or that symptom change is bringing about an improved alliance, rather than the reverse. A recent review which examined eleven studies that met one or both of the criteria suggested by Feeley et al 3 – i.e., alliance predicting subsequent outcome and/or accounting for change that occurred prior to the alliance assessment – reported a median correlation between alliance and outcome of .24, with a mean of .19 4. Although there was some variability in the findings, they supported the conclusion that therapeutic alliance is a key contributor to psychotherapy outcome. However, even such well-controlled studies may underestimate the magnitude of the alliance-outcome relationship. Crits-Christoph et al 5 reported that the use of multiple early alliance assessments, as opposed to just one, revealed a stronger alliance-outcome relationship than is characteristic of current meta-analytic findings; one alliance score accounted for 4.7% of outcome variance, while the average of six early sessions accounted for nearly 15% 5. Patient diagnosis may play an important role in the alliance-outcome relationship. Among well-controlled studies that failed to find a significant relationship between the therapeutic alliance and outcome was one that examined three treatments for cocaine dependence 6. Although absolute alliance levels in this study were reportedly high, it is possible that the “chaotic” nature of the illness in these patients was responsible for the lack of relationship with outcome 6. More recently, researchers have begun to examine the relationship between alliance, pharmacotherapy adherence and outcome. Initial studies suggest that the relationship between alliance quality and symptom outcome is also present in pharmacotherapy. For example, a study controlling for prior symptom change found that early alliance predicted subsequent change in depressive symptoms among those receiving antidepressant medication 7. A probable mechanism for improved outcomes is medication adherence. Indeed, alliance quality has been associated with treatment adherence among those with bipolar disorder 8 and first-episode psychosis, although only provider-rated alliance was predictive in psychosis 9. A smaller literature has examined the relationship between alliance and medication adherence outside the domain of psychiatric prescribing. For example, in patients with systemic lupus erythematosus, higher levels of working alliance were associated with greater medication adherence 10. The relationship between alliance and treatment outcome has stimulated efforts to identify predictors of alliance quality. Few patient-related predictors have been identified. Patient demographic characteristics have not demonstrated consistent relationships with alliance quality. One of the few reliable predictors of high-quality alliances in psychotherapy is a secure patient attachment style 11, which is theorized to develop from predictable and caring interactions with caregivers early in life, eventually manifesting as an enduring relationship pattern. Therapist variability in alliance development appears to have a greater impact on the alliance-outcome relationship than patient variability 12. Therapist attributes such as confidence, warmth, patience and flexibility have been shown to be positively associated with the development of high quality therapeutic alliances 13. Although the alliance is related to nonspecific therapist attributes and skills, it arises in a context in which technical interventions associated with specific forms of treatment are implemented. Considering that the alliance involves agreement on the goals and tasks of treatment, as well as the emotional bond, the contribution of specific techniques to the alliance is not surprising. In a comparison of the relationship between alliance and outcome in a structured form of cognitive therapy versus brief supportive therapy (in which therapists relied primarily on nonspecific elements of treatment), the magnitude of the alliance-outcome relationship was greater in the more structured therapy 14. Additionally, while there were no between-treatment differences in patient ratings of the emotional bond, participants in the more structured treatment rated agreement on tasks and goals more highly than did those in the brief support therapy condition. Thus, different models of intervention are likely to promote the alliance in distinct ways. Data on how to train providers in promoting high-quality therapeutic alliances are few. In one study, trainees were taught specific alliance-promoting procedures including regular collaborative review of goals and therapeutic tasks, and high frequency intentional demonstrations of empathy as well as attention to use of the term “we” during sessions. While the sample was small and the results were not statistically significant, patient-rated alliance from pre- to post-training revealed moderate to large effect sizes 15. The alliance is an ongoing interpersonal process that unfolds over the course of treatment. Some investigators have reported that the alliance manifests a U-shaped or V-shaped course, although the evidence is not consistent. Nonetheless, dealing constructively with alliance perturbations and ruptures is a critical therapeutic task. Recommendations for dealing with such ruptures include openness to criticism, validation of the patient's negative experience, and accepting responsibility for mistakes and empathic failures. Unrepaired ruptures have been shown to predict poor treatment outcome 16. Despite justifiable reservations about methodological flaws in the alliance-psychotherapy outcome relationship, well-controlled studies reveal that alliance is a consistent predictor of outcome. Moreover, the most common alliance measurement approach, the use of single session snapshots, as opposed to multiple early assessments, may underestimate the magnitude of the relationship 5. A nascent literature suggests that the therapeutic alliance is also related to pharmacotherapy adherence and outcome. Despite its consistent relationship with therapeutic outcome, the potential of the alliance has not yet been fully harnessed. Clinicians and those who train them have few guidelines regarding how to create high-quality alliances, and more research is needed in this critical area. Furthermore, although a number of intuitive suggestions for addressing ruptures currently exist, empirical examination of these and other strategies is needed to derive data-based recommendations for responding constructively to alliance perturbations. Finally, given the relationship between alliance and outcome in psychotherapy and more recent similar findings in pharmacotherapy, such research should be extended to areas beyond mental health settings. Although the therapeutic alliance is associated with medication adherence, the potential for employing the alliance to enhance medication adherence outside the domain of psychiatric prescribing has been understudied. Furthermore, treatment approaches for many chronic medical conditions include recommending behavior changes such as exercise or smoking cessation. The role that the patient-provider alliance may play in facilitating changes in these areas has rarely been investigated, but a few studies suggest the patent-rated alliance is related to adherence to treatment recommendations, patient satisfaction, quality of life and self-efficacy for health-related behavior change 17. Hence, re-conceptualizing therapeutic alliance more broadly and studying the impacts of the provider-patient alliance across health care domains may yield insights that improve outcomes for many of the world's most pressing health conditions.