Multiple inclusion criteria were formulated in consultation with all the experts. First, only empirical studies (i.e., consisting of data collection methods) published in peer-reviewed scientific journals, reporting on therapeutic factors and/or mechanisms of change of CATs interventions, were included. Second, we only included articles written in English. Third, an important inclusion criterion was that the study was explicitly situated within a CATs context; in other words, for example, that the CATs intervention was provided by a certified creative arts therapist. Fourth, both quantitative and qualitative data-driven studies as well as systematic literature reviews were included; i.e., that the therapeutic factors and mechanisms of change were derived from empirical data. Case studies were included only when they were data-driven and consisted of empirical methods of data collection. Fifth, because this scoping review focused solely on psychological well-being outcomes, studies that dealt exclusively with medical or physical outcomes were excluded. Finally, no restrictions were placed on client populations so that a wide range of health conditions, gender, culture, context of adversity, and age groups (children, adults, elderly) were included.

Using a pre-defined search strategy, we systematically searched the literature for relevant articles. A computer-based search was conducted of the psychological and medical electronic literature databases without restriction on publication dates, including Medline, Academic Search Complete, Cochrane Library, Web of Science, Embase, Wiley Online Library, Springerlink, PubMed, PiCarta, Academic Search Premier, ScienceDirect, PsycINFO, and Google Scholar. Appropriate search terms were identified based on the findings of the expert group on how “therapeutic factors” and “mechanisms of change” and related terms were defined within the literature. The search strategy included six sets of search terms: those that (i) apply to “mechanisms of change/therapeutic factors,” are relevant to (ii) “creative arts therapies” as an umbrella term, (iii) “art therapy,” (iv) “dance movement therapy,” (v) “drama therapy” or “psychodrama,” (vi) “music therapy.” For all the searches conducted, search terms related to “mechanisms of change/therapeutic factors” were combined with terms referring to “creative arts therapies” or to one of the specific CAT disciplines. See Supplementary Materials for an exemplary search string that was used for the PsycINFO database.

The initial search was conducted by an independent medical librarian and the first and second authors and adhered to the search criteria approved by all the co-authors. All the publications available up to January 2020 that met the inclusion criteria were uploaded to Rayyan QCRI¹, which is a widely used online workspace for screening articles systematically where each screener is blind to the other screeners' selections. This resulted in a total of 4,591 publications, which were then screened in two separate selection steps. All the steps of the selection process were conducted by at least two authors, who selected the studies independently of each other. In cases of selection conflicts, a third author made the final decision. The first selection step, based on title and abstract, resulted in 1,466 publications. The selected publications were labeled according to their specific CATs discipline. In the second selection step, based on full text publications, separate Rayyan workspaces were created for each specific CATs discipline, and for the publications that consisted of more than one CATs discipline. In each workspace, publications were fully screened based on the inclusion criteria by at least two authors specialized in the particular CATs discipline. In addition to the electronic search in the online databases, 11 articles were included based on hand searches of the reference lists in the articles and consultation with experts. In total, 67 articles met all the inclusion criteria for this scoping review, as shown in Figure 1.

To report on the possible and/or assumed therapeutic factors and mechanisms of change in the CATs studies, several characteristics of the included articles were identified and coded independently by two authors using a coding Excel sheet, and conflicts were discussed until an agreement was achieved. In terms of the study characteristics, we coded the design, the type of setting, and context of the study. First, we separated the quantitative studies from the qualitative studies, because the methodology required a different form of coding. The quantitative designs covered randomized controlled trials (RCT), clinical controlled trials (CCT, without randomization), pre-post designs (one group), systematic reviews (SR), or meta-analyses. For each of the quantitative studies, we coded the targeted outcomes, the type of measures used, and the main results. For the qualitative study designs, we coded whether the study used focus groups, interviews, clinical observations, or data from self-report measures. We also coded the research purpose and the main findings of each qualitative study. The type of setting refers to the specific context in which the study was conducted (e.g., schools, forensic psychiatry, palliative care, outpatient- or clinical care). We also coded the target group or client group involved (e.g., children/youth/adults/older adults, inmates, clients suffering from depression or anxiety, older adults suffering from dementia, etc.). We also coded in which country the study was conducted, the frequency and number of therapy sessions, and whether there were group or individual sessions. For the specific characteristics of the studies in each CATs discipline, see Supplementary Tables 1–5.

The first coding step provided initial insights into the articles. Next, we specifically focused on the relationship between the intervention characteristics, therapeutic factors or mechanisms of change, and the effects or targeted outcomes reported in each of the studies. To provide more insights into which of the CATs factors were similar or different from the common factors of psychotherapy, we further analyzed each of the therapeutic factors. First, we coded the therapeutic factors as a common factor (CF) when there was a strong association with the psychotherapeutic factors compiled from the literature on psychotherapeutic common factors (see Table 3). The next step was to identify which of the therapeutic factors that remained could be regarded as a specific factor (SF); i.e., the therapeutic factor was specific to one discipline of the CATs, rather than shared by the CATs disciplines. Therapeutic factors that appeared in more than one CATs discipline, and were not CF of psychotherapy, were therefore regarded as shared factors which we coded as joint factors across the CATs (JF); see Figure 2.

To provide insights into the relationship between all the therapeutic factors from a broader perspective, all the therapeutic factors were categorized by three members of the expert group by inductive analysis. This approach is used in many qualitative data analyses and aims to gain an understanding of meaning in complex data through the development of summary themes or categories from the raw data (e.g., Backett and Davison, 1995; Thomas, 2003). The inductive approach allows research findings to emerge from the frequent, dominant, or significant themes inherent in the raw data, without the restraints imposed by structured methodologies (Thomas, 2006). Starting with the JFs, the inductive analysis further reduced the data into several main categories and corresponding subcategories of therapeutic factors. The second step involved adding the CFs and SFs to these main categories, which resulted in splitting these categories or expanding them with new subcategories. The data analysis process was carried out in its entirety by three researchers and consisted of a continuous iterative process in which all decisions were made by consensus. After the analysis was completed, the whole process was verified by one co-author who was not actively involved in the data analysis. To finalize the results of our analysis we determined whether each domain subcategory was made up of only SFs and JFs of the CATs or only the CF of psychotherapy, or whether it consisted of a mixed therapeutic factor. For a complete overview of the steps taken concerning the data collection and analysis, see Figure 3.

Art therapy takes place in the safety of the studio environment (Van Lith, 2015; Abbing et al., 2018; Nolan, 2019; Bosgraaf et al., 2020) and functions in the context of a strong therapeutic alliance (Bosgraaf et al., 2020; Keidar et al., 2020). Art therapists work in the here and now (Haeyen et al., 2015) to facilitate the expression and clarification of thoughts and feelings visually (Deboys et al., 2017; Abbing et al., 2018; Czamanski-Cohen et al., 2019). Art therapy is motivating (Van Lith, 2015) and offers opportunities to deepen understanding, insight (Haeyen et al., 2015; Van Lith, 2015), and mastery in that art therapy encourages the development of cognitive and emotional skills (Haeyen et al., 2015; Van Lith, 2015; Abbing et al., 2018).

Along with other CATs, art therapy provides a safe and structured pathway for playful, creative experimentation and self-awareness (Deboys et al., 2017; Gabel and Robb, 2017; Abbing et al., 2018; Nolan, 2019). Art therapy, like all CATs, modulates one's sense of time and space; art therapy promotes relaxation and a flow state (Abbing et al., 2018). Individuals and groups benefit from the opportunity to learn and practice artistic skills and forms of visual self-expression (Van Lith, 2015; Bosgraaf et al., 2020). The art-making process helps individuals and groups symbolize and externalize experiences that are not easily verbalized and concretize internal conflicts to facilitate perspective-taking and reflection (Deboys et al., 2017; Gabel and Robb, 2017; Abbing et al., 2018; Czamanski-Cohen et al., 2019; Bosgraaf et al., 2020). This non-verbal expression enables emotional elicitation and processing (Haeyen et al., 2015; Gabel and Robb, 2017).

The tactile quality of art (Abbing et al., 2018) and the choice of appropriate and specific art materials (Haeyen et al., 2015; Bosgraaf et al., 2020) contribute to reducing anxiety and enhancing self-concept. Researchers observe transference during art-making and the artistic product (Hilbuch et al., 2016) as well as the specific potency of art-making in promoting perspective taking and self-awareness. Visual self-expression and the creation of a visual narrative were demonstrated to enhance self-concept, self-esteem and promote emotion regulation (Bosgraaf et al., 2020).

As expected, relational components were important therapeutic factors in DMT studies and were shared across other forms of psychotherapy. The therapeutic alliance (Shim et al., 2017) in one-to-one work and connecting with others (Shim et al., 2017, 2019) in groups were clearly present. Along with external connections, creating connections with one's self was also found to be an important agent of change (Shim et al., 2017, 2019), and finding one's inner balance was presented as another internal therapeutic factor (Mannheim et al., 2013). The DMT studies also highlighted the value of stress reduction (Ho et al., 2018), release and relief through expressing and actualizing emotions (Chyle et al., 2020), but also expanding, regulating, and processing emotions. For example, studies referred to the broadening and building affect, enhancing emotional well-being (Shim et al., 2019), regulating and ultimately transforming emotions (Chyle et al., 2020). Perceiving one's own emotions and blockages (Chyle et al., 2020), along with processes such as reframing (Shim et al., 2017, 2019) or mentalization (Shuper-Engelhard et al., 2019) were all thought to impact therapeutic change. As in other types of psychotherapy, enabling clients to remember important things (Chyle et al., 2020) and facing deeper issues in their past (Winther and Stelter, 2008) was reported to be therapeutic. Finally, the need for a safe environment was seen as an essential requirement for any type of therapy, and more so in DMT where the body and movement may elicit feelings of vulnerability. Creating a safe place was also reported as an important common factor of change within therapy (Shuper-Engelhard et al., 2019).

Common factors in DMT that were present across all the CATs included the discipline's capacity to engage clients in artistic activity and in this case, in movement, through mobilization (Shim et al., 2017, 2019; Chyle et al., 2020). Studies acknowledged that engagement with movement can provide pleasure (Mannheim et al., 2013), similar to artistic pleasure derived from engaging with other art forms. Some found that aesthetics were also linked to therapeutic change as well as the client's experience of beauty (Koch et al., 2016) is key. As is the case for other CATs, creativity and letting go of control (losing control according to Chyle et al., 2020) were regarded as an important first step that can allow for experimentation (designing of model situations and testing alternative behavior, Chyle et al., 2020) and for symbolic work to take place (shifting between the symbolic and the knowing realms, Ellis, 2001). DMT increases brain connectivity (Ventouras et al., 2015), and supports concretization, another important therapeutic process shared across CATs; in DMT this takes place through the direct use of the body and through the therapeutic factor of self-display (Chyle et al., 2020). The re-enactment of biographical or dynamic themes was another therapeutic factor under concretization (Chyle et al., 2020). The ways in which movement experiences are structured within sessions were also viewed as key, since they allow for participation in ritualized processes and the progressive release of emotions (Chyle et al., 2020); thus, supporting the expression of emotions in a safe way.

Embodiment is an important factor of change in DMT practice. Experiencing the body (Chyle et al., 2020) and body awareness (Mannheim et al., 2013; Shim, 2015) are regarded as an important agent of change. Similarly, although verbalizations are not excluded, the non-verbal nature of DMT allows for discharge of energy and release of tension (Mannheim et al., 2013), expressivity in movement (Shim, 2015) and non-verbal disclosure (Chyle et al., 2020). This engagement with movement can communicate unconscious processes in the form of movement metaphors (Ellis, 2001) and support insight by relying on bodily sensations as a source of information (Chyle et al., 2020). It can also support the development of artistic skills such as connecting movement and language with one another or learning and practicing motion sequences (Chyle et al., 2020). Artistic agency is also facilitated and can be evident in the ways in which one moves. Koch et al. (2007), for example, showed that the jumping rhythm is presented as an important agent of change in clients with depression. Finally, the therapeutic alliance is encouraged through kinesthetic empathy in the form of mirroring (Shuper-Engelhard et al., 2019), dialogue in the form of movement interactions and synchrony in the form of movement synchronicity (Chyle et al., 2020).

Drama therapy and psychodrama employ common factors of psychotherapy such as the importance of establishing a safe environment and the value of a positive therapeutic alliance (Cassidy et al., 2014; Orkibi et al., 2017a). Drama therapists and psychodramatists employ techniques that facilitate experiencing in therapy, to develop a felt awareness (Armstrong et al., 2016). Drama therapy and psychodrama promote a sense of agency and empowerment (Bucută et al., 2018), hope and optimism (Bucută et al., 2018), reflection (Cassidy et al., 2017), self-awareness and insight (Bucută et al., 2018; Testoni et al., 2018) and facilitate emotional release and relief (Bucută et al., 2018). Four psychodrama studies examined the transformative potential of group processes, such as Yalom's common factors for group therapy, with a specific emphasis on catharsis (emotional release), insight, self-understanding, interpersonal learning, and the reciprocal process of giving and receiving support (Yalom, 1983, 1995; Kellermann, 1985, 1987; Oezbay et al., 1993; Kim, 2003; Testoni et al., 2018). Two studies dealt with the ways in which the observed changes were related to non-specific and productive in-session client behaviors such as cognitive, behavioral, and emotional exploration (Orkibi et al., 2017a,b) and one study highlighted the catalytic function of working in the here and now (Cassidy et al., 2014).

Drama therapy and psychodrama, like all CATs, elicit active involvement (Cassidy et al., 2014) within an embodied artistic process in which both verbal and non-verbal expressions are possible (Cassidy et al., 2017). Like other CATs, drama therapists and psychodramatists offer intrinsically pleasurable and playful (Orkibi et al., 2014) activities that promote a sense of agency in that participants experience control and choice, and creative experimentation where clients are invited to practice their spontaneity and try out new ways of being (Orkibi et al., 2014; Cassidy et al., 2017). Drama therapists and psychodramatists, like other CATs, employ concretization, the process of rendering internal experience visible and tangible, which further enables perspective-taking and insight (Cassidy et al., 2017). Witnessing, a reflective process common to all CATs, was also highlighted in one study (Orkibi et al., 2014).

Drama therapists and psychodramatists facilitate change through active engagement with and within dramatic or surplus reality (Cassidy et al., 2014; Orkibi et al., 2014, 2017a). Emotional regulation and reflection are facilitated by working at a safe distance within or outside the drama (Cassidy et al., 2017). Drama therapists and psychodramatists encourage understanding, self-awareness, perspective, and empathy through doubling (Goldstein, 1971), role-reconstruction (Bucută et al., 2018), encounter, and role-reversal (Orkibi et al., 2014; Testoni et al., 2018). In addition, dramatic embodiment and dramatic projection were identified as two specific factors contributing to client experiencing a felt awareness, which facilitates change (Armstrong et al., 2016).

The common psychotherapy factors in music therapy include working in the here and now (Ansdell et al., 2010; Carr et al., 2012) within a positive therapeutic alliance (Bonde, 2005; Kellet et al., 2019) in a safe, predictable environment (Robb, 2000; De Witte et al., 2020c). Several studies emphasized the importance of group processes such as group cohesiveness (Goldberg et al., 1988; Waldon, 2001; Ahonen-Eerikäinen et al., 2007; Carr et al., 2012; Dalton and Krout, 2015; Bibb and McFerran, 2018; De Witte et al., 2020a), feelings of togetherness and bonding (De Witte et al., 2020b), altruism (Gardstrom et al., 2017), validating feedback by others, and interpersonal learning (Goldberg et al., 1988; Gardstrom et al., 2017). Participants in music therapy experience a sense of personal connection (Rolvsjord, 2010; Dalton and Krout, 2015), pro-social skills (Warth et al., 2016), meaning (McDermott et al., 2013; Baker et al., 2015), agency (Ahonen-Eerikäinen et al., 2007; Potvin et al., 2018), motivation (Dalton and Krout, 2015; Gardstrom et al., 2017; Millet and Gooding, 2017), and emotional release and relief (Gardstrom et al., 2017). Client in-session behavior factors of involvement and engagement were represented in all studies.

Music therapy, like all CATs, actively engages participants (Landis-Shack et al., 2017; Millet and Gooding, 2017) in a creative process (Rolvsjord, 2010) where verbal, symbolic (Bonde, 2005; Short et al., 2009; Baker et al., 2015; Bibb and McFerran, 2018), and non-verbal expression are possible (Ansdell et al., 2010). Music therapy heightens participants' sense of artistic agency (Potvin et al., 2018) and promotes emotional release and relief through deep relaxation (Short et al., 2009). Like all CATs, music therapy promotes well-being because it is intrinsically enjoyable (Ahonen-Eerikäinen et al., 2007; Gardstrom et al., 2017), and because it modulates one's sense of time and space, by bringing one into a state of flow (Baker et al., 2015) or as a distraction from stress-inducing thoughts (Porter et al., 2017).

The specific therapeutic factors in music therapy include the physical act of music making (Landis-Shack et al., 2017) and the safe and structuring (Robb, 2000) nature of music itself including the use of tone, tempo, and alternating rhythms (Ansdell et al., 2010; De Witte et al., 2020a,b,c). Music therapy enhances therapeutic alliance and group processes through playful (Passiali, 2012; Porter et al., 2017) musical interactions (Passiali, 2012; Bibb and McFerran, 2018), shared musical experiences (Porter et al., 2017), musical attunement (McDermott et al., 2013), musical synchronicity (Potvin et al., 2018), and musical dialogue (Kellet et al., 2019). Music therapy triggers or activates memories (McDermott et al., 2013; Porter et al., 2017; Bibb and McFerran, 2018) which, together with intervention choices and musical cues facilitate coping and emotion regulation (Robb, 2000; Baker et al., 2015; Landis-Shack et al., 2017).

These studies confirmed the impact of several common factors of change in psychotherapy beginning with the therapeutic alliance (Heynen et al., 2017). Their orientations differ in that some work in the here and now (Flanagan, 2004; Parsons et al., 2020), while others focus on past events (Parsons et al., 2020), creative arts therapists tailor the therapeutic goals to meet individual and group needs, facilitate fundamental relational skills and features, and promote supportive relationships (Heynen et al., 2017; Kalaf and Plante, 2019; Parsons et al., 2020). CATs take place in a liberating environment characterized by trust and safety (Chiang et al., 2019), in which clients can experience a sense of agency and skill with regard to processing and communicating emotions and forming connections with others (Dunphy et al., 2019; Kalaf and Plante, 2019). Creative arts therapists facilitate self-awareness and opportunities for meaning-making (Flanagan, 2004; Kalaf and Plante, 2019).

These studies revealed that the CATs actively engage (Parsons et al., 2020) participants in artistic therapeutic activity and offer multiple options for verbal and non-verbal expression through symbolism and metaphor (Flanagan, 2004; Schiltz, 2014; Chiang et al., 2019; Dunphy et al., 2019). The CATs facilitate concretization, to make visible internal conflicts, which enables perspective-taking (Schiltz, 2014; Kalaf and Plante, 2019). In addition, participants experience a sense of creativity, artistic pleasure, and achievement as they develop artistic skills (Chiang et al., 2019; Dunphy et al., 2019; Parsons et al., 2020).

The findings showed that specific factors are related to artwork and art-making such as promoting positive therapeutic change by seeing one's own emotions through the artwork, eliciting a “visual narrative of life” and “portraying the feelings of past/future” (Haeyen et al., 2015; Hilbuch et al., 2016; Bosgraaf et al., 2020). The artwork, in this way, becomes a tangible and concrete agent that facilitates perspective taking, motivates self-reflection and self-awareness, and further promotes understanding during the therapeutic process (Abbing et al., 2018; Bosgraaf et al., 2020). This finding aligns with the recent Adaptive Response Theory art therapy framework, which situates art-making as well as the art product as distinct elements that enable clients to shift from maladaptive to adaptive responses through the interpersonal and intrapersonal understanding that occurs in the session (Kaimal, 2019).

Another important domain of specific factors in art therapy is interaction and engagement through the physicality of the various art materials. Discovering materials and possibilities with art media (Haeyen et al., 2015) have been associated with a sense of mastery, embodied imagination, and artistic agency. The tactile quality of the art media (Abbing et al., 2018) and the use of specific art materials and techniques (Bosgraaf et al., 2020) facilitates exploration and creative engagement during the art therapy process. This is consistent with the theoretical model of engagement with the art media at various kinesthetic/sensory, perceptual/affective, and cognitive/symbolic levels that are all linked at the creative level (Lusebrink et al., 2013). However, the specific mechanisms by which the different dimensions of the artistic medium facilitate various levels of engagement deserve further theorization and empirical investigation.

The non-verbal nature of DMT is considered an important therapeutic factor (Mannheim et al., 2013; Shim, 2015; Chyle et al., 2020). As the definition of the discipline suggests (Karkou and Sanderson, 2006), attention to the body and movement allows for body-mind connections and for holistic attention to therapeutic change. Similarly, embodiment, in the form of experiencing the body (Chyle et al., 2020) and body awareness (Mannheim et al., 2013; Shim, 2015), is an important factor of change in the DMT studies included here. This finding echoes theoretical discussions on embodiment in the discipline (Koch and Fuchs, 2011; Fuchs and Koch, 2014; Payne et al., 2019) as well as applications of these ideas to clinical practice (Sandel et al., 1993; Pallaro, 2007) and research (Meekums et al., 2015; Pylvänäinen et al., 2015; Pylvänäinen, 2018; Karkou et al., 2019). Mary Whitehouse (Pallaro, 2007) for example, one of the early pioneers of DMT in the USA, and a Jungian analyst, highlighted the value of heightening sensorial messages as a way to provide information in the here-and-now, but also as a way to allow one to delve into the depth of one's inner world and past history. Similarly, Ellis' (2001) discussion of movement metaphors refers directly to revealing unconscious processes, and Chyle et al's (2020) references to bodily sensations are regarded as supporting insight. Both findings echo influences of Jungian thinking and depth psychology in DMT practice and support arguments that the discipline is a form of creative psychotherapy where movement metaphors play a central role (e.g., Meekums, 2002).

Influences from movement analysis systems and movement-based practices such as Laban (McCaw, 2012) and Kestenberg Amighi et al. (2018) can also be identified in the specific DMT factors. Chyle et al. (2020), for example, noted the development of artistic skills including connecting movement with language and learning certain movement sequences, while Koch et al. (2007) explored the role of the jumping rhythm for developing artistic and general agency as described by Kestenberg Amighi et al. (2018).

Finally, mirroring (Shuper-Engelhard et al., 2019), movement interactions and movement synchronicity (Chyle et al., 2020) all emerged as important therapeutic factors. The development of the therapeutic alliance, dialogue and synchrony are part of the fabric of DMT practice, and were strongly advocated by Marian Chace and her students and explicitly flagged up by Schmais (1985) in one of the first texts on therapeutic factors written in the discipline. Schmais (1985), in this seminal text, talked about healing processes in DMT, translating Yalom's therapeutic factors in verbal psychotherapy groups into DMT group practice. The assumption that these are important and unique ingredients of DMT that are responsible for therapeutic change remains largely the same. Conversely, the contemporary literature translates practice-based theorization into empirical investigations, as evidenced by the DMT publications included in this review.

Overall, we identified more PD than DT specific factors. Beyond the fact that there were more PD than DT studies included in this review, PD is more structured and unified than DT in terms of both theory and practice (Kedem-Tahar and Kellermann, 1996). Therefore, the therapeutic factors in PD are relatively well-defined and are often quantitatively measured through observational ratings or self-reports. As such, they lend themselves more readily to consistent operationalization and empirical investigation. For instance, doubling and role reversal are well-defined PD techniques, while encounter, role-reconstruction, and catharsis are well-defined key factors in the PD theory of change (Moreno, 1972/1994; Blatner, 2000; Orkibi, 2019). PD therapeutic factors that call for further empirical investigation include spontaneity and creativity, concretization, act hunger, act fulfillment, and action insight, to name a few.

In contrast, the DT therapeutic factors of dramatic projection and embodiment, which were identified in qualitative studies, need further operationalization before they can be quantitatively measured in future change process or process-outcome studies. For example, Jones (2007) suggested the following eight core processes that apply to all DT approaches and reflect in-session drama and theater therapeutic processes: dramatic projection, drama therapeutic empathy and distancing, role-playing and personification, interactive audience and witnessing, embodiment, playing, the life-drama connection, and transformation (pp. 99–129). Finally, further operationalization of Landy's (1997) concept of aesthetic distance may provide a discipline-specific conceptualization for clients' levels of emotional involvement in a DT session that may be associated with a range of meaningful psychological outcomes such as emotion regulation.

The findings show that specific factors related to the musical dialogue, such as “shared musical experiences” or “musical interactions,” are very often associated with positive therapeutic change. This suggests that the therapeutic relationship itself is formed in the music. By offering a solid musical frame, any musical expression produced by the client can be musically encouraged and responded to in a musical dialogue (Nordoff and Robbins, 1966; Aigen, 2005; MacDonald et al., 2013). The 64 musical improvisation techniques developed by Bruscia (1987) based on using the unique qualities of music to establish/influence the musical dialogue, are still the basis of music therapy training all over the world. However, since musical dialogue contains so many different components and working processes, both inside and outside the music, it is strongly recommended to conduct further research into related therapeutic factors, for example through micro-analysis of a short segment of a session (Lee, 2000; Wosch and Wigram, 2007).

Another important domain of specific factors in music therapy associated with a wide range of positive outcomes (e.g., reduction of stress, arousal, anxiety), concerns the structuring nature of music, which were referred to in the studies as “repetitive rhythms,” “tempo of the music,” and “musical simplicity.” Neurological evidence shows that music with a slow and steady rhythm can provide relaxation and calm by altering autonomic body rhythms such as heart rate (e.g., Thaut et al., 1999, 2015; Bernardi et al., 2006; Thaut and Hoemberg, 2014; Koelsch, 2015). This is in line with two recent meta-analyses which reported positive correlations between a slow, steady music tempo and stress/anxiety reduction (De Witte et al., 2020a,b). However, most of these studies used pre-recorded music and music tempo was usually not measured during a music therapy session using improvised live music. Therefore, we recommend further investigation of music tempo as a specific therapeutic factor within active music therapy interventions, as related to psychophysiological and psychological outcomes.

This scoping review underscores the complexity of studying therapeutic factors. Overall, there is a need for a more unified vocabulary and a greater specificity in conceptualizing and operationalizing therapeutic factors in CATs. One necessary starting point is a shared terminology (i.e., Table 2) used in an accurate, systematic, and consistent way by the CATs research community. Shared definitions help recognize and minimize redundancy, confusion, and contraindications. For example, in some studies referring to a measured outcome variable (e.g., enhanced self-concept, emotion regulation, empowerment, stress) as a therapeutic factor is questionable given the fact that such variables are more often than not defined in the psychology and psychotherapy literature as outcomes rather than therapeutic factors. Nevertheless, a compelling theoretical argument for conceptualizing such outcome variables as therapeutic factors was also often found to be lacking.

In line with Kazdin (2007, 2009), it is crucial to use theory as a guide to explain how a therapeutic factor operates across CATs and in a given discipline. A theory can be studied bottom-up (by exploratory qualitative or quantitative data) and/or top-down (by an existing theory of change). For example, Blatner (1992) described the underlying therapeutic principles of all CATs and Johnson (1998) introduced a psychodynamic model that explains CATs' therapeutic action. More recently, Koch (2017) suggested the framework of embodied aesthetics as a model for specific mechanisms in CATs. There are also discipline-specific models of therapeutic factors; for example, in drama therapy (Jones, 2007), psychodrama (Kushnir and Orkibi, 2021), art therapy (Czamanski-Cohen and Weihs, 2016; Kaimal, 2019), dance movement therapy (Koch and Fischman, 2011; Imus, 2021), and music therapy (Baker and Roth, 2004; Grocke, 2019).

To examine how change occurs, psychotherapy researchers often look at mediators, but this line of studies is generally lacking in CATs studies and therefore calls for further discussion here. A mediator (M) is an intervening variable that is theorized to account for the causal statistical relationship between two variables, such that X causes M which in turn causes Y (see Figure 4A). In psychotherapy research, a mediator is situated between the intervention (independent variable, X) and the outcome (dependent variable, Y) (Kazdin, 2009). For example, a psychodrama intervention (X) may increase perspective taking (Y) through role reversal (M). It is possible to examine a more complex theory of change that has parallel mediators or a chain of sequential mediators in a single statistical model, as illustrated in Figures 4B,C (see also Hayes, 2013). Thus, we propose reserving the term mediator or mediating variable for the statistical relationship between variables, in the context of statistical mediation analysis.

However, “mediation does not equal mechanism” (Nock, 2007, p. 5S), because isolating a mediator does not itself explain how treatment leads to change. Kazdin (2009) noted that while the study of mediation is the first step in understanding how therapy works, “mechanism refers to a greater level of specificity than a mediator and reflects the steps or processes through which therapy (or some independent variable) actually unfolds and produces the change” (p. 419). Proposing a mechanism of change is first and foremost a theoretical task that should reflect the theory of change posited by the researcher. This includes a causal sequence of “events” (or mediators in quantitative mediation analysis) that reflect how the change occurred. To illustrate, a strong client-therapist alliance (a common change factor serving here as an independent variable), may lead to more dramatic engagement in drama (a specific change factor, M1) that in turn leads to more action insight (another specific change factor, M2), which enhances self-understanding (a common change factor, M3), and consequently to behavioral change (the dependent variable). This example illustrates a causal process (i.e., sequential chain of variables) reflecting one possible mechanism of change in psychodrama that involves both common and specific factors. Note that the independent variable can also be the intervention (e.g., psychodrama vs. CBT) such that the treatment variable precedes the alliance. Mediators should be clearly articulated in the theoretical model upon which the treatment is based in order to establish “treatment differentiation” that clarifies the essential factors by which a given intervention differs from a comparison or control group (Ang et al., 2018). Creative self-efficacy is a particularly relevant construct that offers promising avenues for future CATs research on mechanisms, because it has been recently associated with creative adaptability and well-being indicators (Orkibi, 2021).

A moderator is a construct external to the treatment that influences the direction or magnitude of the relationship between the treatment and outcome (see Figure 4D). Hypothesizing moderators can clarify when or for whom therapy leads to change, because moderators are often client or therapist characteristics (e.g., gender, ethnicity, years of practice, temperament) or the treatment delivery format (e.g., individual vs. group; in-person vs. online) and dose (e.g., once vs. twice a week) (Kazdin, 2009). It is noteworthy that in some cases “a given variable may function as either a moderator or a mediator, depending on the theory being tested” (Frazier et al., 2004, p. 116). More complex statistical analyses can examine a moderated mediation model in which “the effect of X on Y through M is moderated by or conditioned on one or more variables” (Hayes, 2013, p. xi). In other words, the mediation “operates to varying degrees (or not at all) for certain people or in certain contexts” (Hayes, 2013, p. 432). In a mediated moderation model, “the interaction between two variables affects a mediator, which then affects a dependent variable” (Morgan-Lopez and MacKinnon, 2006, p. 77). However, it has been claimed that the latter model “rarely has a meaningful substantive interpretation” and thus should be avoided (see Hayes, 2013).

Note that while qualitative research does not involve statistical analysis of possible causal relationships between variables, empirical qualitative research draws on participants' and/or the therapists' self-reports of their treatment experiences to describe possible therapeutic factors, or more detailed mechanisms of change, and vice-versa. Relatedly, labeling the therapeutic factors identified in a qualitative study as mediators should be avoided, since this term is conventionally used in the context of quantitative statistical analysis.

We call on CATs researchers to include two or more measures of candidate mediators in quantitative treatment studies in order to identify which mediator makes a greater contribution to the outcome. Mediators may be related to the therapist, the client, the interaction, and/or the group. However, to demonstrate a causal timeline (i.e., temporal sequence) it is important to assess both the mediator and the outcome (preferably in multiple occasions across the treatment), “to ensure the mediator has, in fact, changed before the outcome” (Kazdin, 2009, p. 424). Both control studies (in a laboratory) and naturalistic studies (in a real-world setting) are valuable to identify mediators and therapeutic factors. In controlled studies, the direct experimental manipulation of a proposed mediator can provide evidence that the change in the outcome is a function of the levels of the manipulated mediator (e.g., a group with low vs. a group with high dramatic engagement). Naturalistic studies can be useful as well for generating and testing hypotheses about therapeutic factors in treatments delivered in the field (Kazdin, 2009). The conditions under which a therapeutic factor impacts a given outcome (i.e., when or for whom) can be assessed with a conditional meditation analysis (i.e., “moderated mediation”) with the mediator being specified, for example, as a dose (e.g., treatment length), different client characteristics (e.g., age, gender), and varying settings (e.g., individual, group).

Moreover, the CATs field can progress through research that pinpoints therapeutic factors in the artistic and creative process, thus pointing to CATs as treatment methods in their own right. To this end, we urge researchers to develop methodologically rigorous measures that capture CATs' unique therapeutic factors, including those designed to determine the embodied and non-verbal expressive and creative processes that can be assessed by means of triangulation to strengthen internal validity (e.g., self-reported, therapist reported, and observational data). These measures should be disseminated to the CATs research community, and in particular to faculty members who are active researchers in research-oriented CATs programs, who may help establish their validity, reliability, and replicability across settings and populations.

Finally, the development and testing of CAT micro-interventions responds to the call for more research on therapeutic factors. Micro-interventions can be regarded as a short part of a session in which the therapist uses specific therapeutic techniques or steps to work on the client's specific goals (De Witte et al., submitted), which is in line with the growing recognition that one-size-fits-all approaches to intervention may be suboptimal for the client and healthcare system alike (Rush et al., 2004; Gauthier et al., 2017). Developing CAT micro-interventions may be a suitable way to uncover a therapist's implicit and practice-based knowledge, which may stimulate the transferability of valuable clinical practices and provide more insights into which factors lead to change. Micro-intervention testing also allows researchers to conduct micro-analyses on a very short segment of the therapeutic process and is therefore highly suitable for pinpointing elements of the intervention that cannot be examined when testing over a larger period of time. Ultimately, such knowledge exchange will promote the consistency and replication of results across independent studies, thus enhancing possibilities to draw substantiated inferences about a given therapeutic factor (Kazdin, 2009).