The Therapeutic Alliance in Psychedelic-Assisted Psychotherapy: A Novel Target for Research and Interventions

Abstract

Clinical trials support the promise of psychedelic-assisted therapy (PAT) and reveal potential mechanisms underlying efficacy. Out of necessity, initial studies of PAT have focused on feasibility and impact with special attention to acute responses to the molecules, but investigations of the psychotherapy components of PAT have lagged other work. Years of research on other forms of psychotherapy reveal that one of the most important contributors to outcome is the therapeutic alliance, a cooperative connection between clients and providers. The alliance has accounted for meaningful variance in psychotherapies independent of their theoretical orientation, number of sessions, and rates of change in improvement. Nevertheless, recent critiques suggest that previous estimates of the alliance's impact were exaggerated because of statistical problems. This controversy only underscores the idea that the alliance could prove a valuable target for understanding mechanisms and enhancing PAT's impact. Initial work shows promise, but only two publications address the issue directly despite decades of emphasis on the alliance in published recommendations for conducting PAT. Adding alliance measures to clinical trials would not require extensive effort or resources. Each practitioner could improve alliance relatively easily, regardless of theoretical orientation, through increases in uncontroversial clinician behaviors that focus on respect for client autonomy, listening skills, and current practical concerns. This article details support for the alliance construct, reviews relevant measurement strategies, and underscores implications for researchers and clinicians. Increased attention to the therapeutic alliance could enhance our understanding of mechanisms underlying PAT and improve outcomes, potentially decreasing human suffering more efficiently.

The Promise of Psychedelic-Assisted Therapy

Psychedelics combined with supportive therapy have promoted rapid sustained improvements in individuals with many psychiatric conditions, including major depressive disorder,¹ post-traumatic stress disorder,² and substance use disorders.³ In response to favorable attitudes toward psychedelic-assisted therapy (PAT), venture capitalists have invested billions of dollars in anticipation of psychedelic decriminalization.⁴ Media coverage contributes to this fanfare; stories tout psychedelics as “miracle medicines” capable of generating remarkable cures.⁵

Despite cautions about methodological limitations in available data,⁶ current optimism has inspired more research and new applications. Proposed replications and extensions will require reasonable expectations and appropriate interpretations,⁷ but these trials also present an opportunity to clarify our understanding of underlying therapeutic mechanisms.

Mysteries of Mechanisms

Mechanisms underlying PAT-induced improvement remain unclear. Research often focuses on psychedelic components, linking therapeutic outcomes to physiology⁸ or subjective effects of psychedelics, including mystical experiences or emotional breakthroughs.^9–11 The psychotherapy component has generated less attention, though theorists emphasize nondrug contextual contributors to long-lasting benefits,^12–15 and nondrug sessions represent the bulk of PAT.¹² Decades of theorizing stress that fostering optimal set and setting facilitates positive subjective responses to psychedelics that subsequently lead to improvements.^16,17

Accordingly, recommendations emphasize environmental aspects (preparation and integration visits, therapist stance during dosing, music, etc.) in part because of their potential impact on subjective effects. Examinations of other facets of the treatment seem appropriate.

Little empirical research examines the role of the psychotherapy component in PAT. Theorists propose that enhancements stem from cognitive or meta-cognitive changes in constructs such as psychological insights, flexibility, or dysfunctional attitudes, which only some therapists might find familiar.^10,18 Research on psychotherapy without psychedelics suggests that one construct accounts for variance in outcomes regardless of treatment type: the therapeutic alliance.¹⁹ Authors define this alliance as the collaborative bond between patient and clinician that reflects mutual engagement in treatment.¹⁹

After a century of theorizing and decades of research, this alliance remains the one mutually agreed upon hypothesized therapeutic mechanism.²⁰ Effects of the alliance exceed those of treatment duration, rates of change in improvement, early perceptions of credibility of the therapy,²¹ or the specific type of intervention.^22,23 Even proposed mediators can benefit from a good therapeutic alliance.¹⁹ Therapeutic alliance has the potential to affect outcome both directly and indirectly.

But even this widely accepted contributor eventually drew criticism. Recent work on links between alliance and outcome emphasizes that multiple factors can exaggerate estimates of their covariation. The reciprocal associations between alliance and outcome, the causal impact of either on itself later, or correlated errors of measurement across time inflate estimates. Statistically detrended approaches suggest that the association is modest at best.²⁴ Nevertheless, others viewed this modest estimate as incorrect after disentangling alliance as a dynamic construct with a “state” and a “trait” component.²⁰

This dynamic view suggests that qualities in both client and therapist, independent of their relationship, contribute to the trait component, but varying facets unique to their interactions create the state component. Both play a role in the outcomes of standard psychotherapy, and likely account for important variance in PAT's effects, as theorists have emphasized.¹³ Given established links between the therapeutic alliance and a therapist's perceived social support and comfort with closeness in relationships, personal therapy could enhance the trait components as well.²⁵

The Impact of Therapeutic Alliance

Alliance and outcomes improve together according to multiple meta-analyses, each reviewing at least 100 studies. Note that Pearson's R (the correlation between alliance and outcome) and k (the number of studies providing estimates) reveal consistent support: Flückiger et al.²⁶: r = 0.28, k = 295; Horvath and Bedi²⁷: r = 0.21, k = 100; Horvath et al.²⁸: r = 0.28, k = 190; Horvath and Symonds²⁹: r = 0.28; k = 190; Martin et al.²³: r = 0.22, k = 79. Reviews include overlapping studies; an average across them might prove misleading. Nevertheless, estimates appear stable. The implications of effects of this size can seem difficult to envision, even for statisticians.

An alternative approach to interpretation, dubbed the Binomial Effect Size Display,³⁰ can prove illustrative. Dichotomizing alliance as stronger versus weaker and outcomes as improved versus not, and accepting a few assumptions, can aid understanding. The smallest estimate (r = 0.21) suggests that the strong alliance group would have 60.5% of clients improve, whereas only 39.5% of the weak alliance group would improve. The largest estimates (r = 0.28) suggest even more disparate outcomes (64% vs. 34%). Thus, a strong alliance (or relevant correlates) could increase chances of improvement up to 30 percentage points. Researchers, clinicians, and clients could find an effect this large quite meaningful.

Although these meta-analytic estimates do not include PAT research, preliminary work supports the alliance's impact for this treatment. Nevertheless, the number of relevant data-based publications seems limited to only two experiments. In a randomized controlled trial investigating psilocybin as a treatment for depression, a strong therapeutic alliance for those in the PAT arm predicted more emotional breakthrough and mystical experiences, leading to improvements with an effect larger than those reported in meta-analyses (r = 0.49).³¹ In contrast, the strength of the alliance did not predict depression in the escitalopram comparator arm of the same experiment.

This result underscores the import of therapeutic alliance for PAT, especially relative to commonly prescribed antidepressants. In a second publication examining the impact of psilocybin and meditation on well-being, more therapist support (experimentally manipulated) facilitated better outcomes. Those in the high-support condition reported more positive daily experiences, gratitude, and sense of purpose.³² These data suggest that therapeutic alliance might play a role in PAT as important as its role in other psychotherapies, as investigators emphasize.^13,14 In addition, this alliance is arguably critical for any of the therapeutic approaches applied in PAT,^12,33 which can include standard cognitive behavioral interventions as well as other therapies.³⁴

The strength of the alliance likely depends upon multiple facets of PAT, including the therapy employed,³⁵ support during acute administration,³⁶ and therapist and client characteristics,³⁶ but clearly has considerable potential worthy of investigation and application. Addressing not only positive but also negative aspects of the relationship will be critical for researchers and clinicians. In addition to its ubiquitous direct effects, under special circumstances, the alliance can also lead to better outcomes as it deteriorates. For example, when the quality of object relations is low, an alliance that diminishes over time can lead to greater improvement.³⁷ The alliance might also function as a moderator of intuitive links between other therapist variables and outcome, revealing associations in its presence that might fail to reach significance in its absence.

For example, competence shows a positive correlation with outcome in the presence of a strong alliance in cognitive processing therapy for post-traumatic stress disorder, but essentially no association when the alliance is weak.³⁸ The alliance might show particular promise given the novel format of PAT relative to other psychotherapies (e.g., longer sessions, multiple therapists present during dosing, and level of care).¹² Debates about the alliance's state and trait components (as described in the “mysteries of mechanisms” section above) underscore the need to examine the construct in PAT.^20,24

Implications for Researchers

Given the potential impact of the therapeutic alliance, PAT researchers could account for important variance in outcomes, and illuminate underlying mechanisms, by measuring this construct at opportune times. The measurement need not be expensive, effortful, or aversive, especially to illustrate initial promise. Although researchers debate the optimal measure of therapeutic alliance, most scales show comparable links with improvements. Any accepted approach could prove informative.^39,40

Ideally, the therapist, client, and a rater who watches each session (blind to condition) should show enough range in method variance to create trustworthy indices of strength. Even simple self-reports from client and therapist would inform proposed trials. Common measures include The Working Alliance Inventory and The Barrett–Lennard Relationship Inventory, both of which predict therapeutic outcomes in other work.⁴¹

The Working Alliance Inventory has an acceptable 12-item short form, good psychometric properties, and evidence of validity. The scale focuses on both relationships and processes. Typical items have considerable face validity including, “As a result of these sessions I am clearer as to how I might be able to change” and “I believe my therapist likes me.”⁴² The Barrett–Lennard Relationship Inventory draws from Rogerian theory and focuses on relationships.

Psychometric properties appear sound; initial steps on the development of a 12-item short form appear promising.⁴³ Items include “I believe my therapist usually understands the whole of what I mean” and “My therapist feels affection for me.”⁴⁴ The empathy subscale correlates (r = 0.30) with the Working Alliance Inventory.⁴⁵ Perhaps these scales measure comparable constructs. Either (or both) could add to the PAT literature.

The timing and frequency of assessments will likely vary with treatment plans and resources. Most PATs include preparatory, dosing, and integration sessions.⁹ Assessing at each stage has intuitive appeal. Adapting versions of scales for therapists, guides (the sitters during drug administration), clients, and observers appears straightforward. Correlations among rater reports are often low; a therapist, client, and independent observer might provide very disparate scores on items that assess the alliance.⁴¹ Emphasizing lowest scores might reveal a threshold for optimal outcomes or “readiness” for acute dosing. Disparities between ratings might address the role of varied perceptions on progress.

An assessment at every session seems defensible, but reactivity and rater burden might suggest lower frequencies. Given the novelty of these assessments in PAT, even a single observation could enhance investigations, though the best work measured the construct repeatedly across treatment sessions.³¹ Multiple ratings across multiple sessions would have advantages, especially since lone pre- and postmeasures would ignore the nonlinear nature of alliance building.^20,24,46 Researchers could add measures to protocols without outrageous expense or effort. Doing so can illuminate underlying variables to target during therapist training (e.g., rupture repair strategies) while shedding light on differences in within-therapist and between-therapist alliance correlations.⁴⁷

Implications for Providers

Given the import of the alliance in psychotherapy^19,39 and its promising role in PAT,^31,32 practitioners might improve outcomes through this mechanism. Enhancing alliance seems more straightforward than altering other established predictors of PAT outcomes. Correlates of a strong therapeutic alliance include supportive actions within the repertoire of trained clinicians. Correlates include behaviors consistent with universal definitions of good care, such as expressing validation, empathy, interest, respect, and belief in client strengths.

A stronger alliance appears when therapists underscore progress, establish goal consensus, remain flexible and responsive to client needs, focus on the present, and concentrate on practical issues in supportive/expressive dynamic therapy.⁴⁸ Other actions might be more relevant to the alliance in other treatments. Breaking these broad categories into individual techniques can miss the spirit underlying a strong working alliance, which benefits from treating clients as partners, respecting autonomy, listening well, and asking permission before providing input.⁴⁹

Clinicians tend to believe they are adhering to protocols; they validate while showing empathy, respect, interest, and optimism. Nevertheless, therapists might drift from protocols more than they realize, increasing the use of modifications to therapy that are not empirically supported.⁵⁰ Relevant reminders, regular supervision, and supportive discussions among clinicians can certainly help.

Conclusions

Although large meta-analyses suggest that a strong therapeutic alliance improves psychotherapy, recent controversy raises questions about the magnitude of the relationship's impact.^20,24 Over 7,000 articles published since 2019 on Google Scholar mention “psychedelics” and “therapeutic alliance.” Theorists consistently stress the alliance's import in discussions of PAT. Some suggest that psychedelics might create improvements through the enhanced relationship with the clinician.^12–14 Nevertheless, empirical work seems limited to two articles.^31,32

Both publications reveal promise, but the clarion call for continued research seems relevant. The current article emphasizes the importance of measuring this construct regardless of treatment approach, identifies scales and options for assessment, and details correlates that clinicians currently engaged in PAT might relish enhancing. These efforts need not require massive resources; they would have tremendous potential for revealing mechanisms and enhancing results. As data and clinical experience amass, these inexpensive steps can improve outcomes and our understanding of PAT so that relevant work can alleviate suffering.

Author's Contributions

P.K.-B. contributed to conceptualization, writing—original draft, and writing—review and editing. E.B.G. was involved in writing—review and editing. M.E. contributed to conceptualization and writing—review and editing.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.