Implementation of a biopsychosocial approach into physiotherapists’ practice: a review of systematic reviews to map barriers and facilitators and identify specific behavior change techniques

Abstract Purpose Our first objective was to map the barriers and facilitators to the implementation of a biopsychosocial approach into physiotherapists’ practice within the Theoretical Domains Framework (TDF). Our second objective was to identify the specific behavior change techniques (BCT) that could facilitate this implementation. Materials and methods We conducted a review of systematic reviews to identify barriers and facilitators to the use of a biopsychosocial approach by physiotherapists and we mapped them within the TDF domains. We then analyzed these domains using the Theory and Techniques tool (TaTT) to identify the most appropriate BCTs for the implementation of a biopsychosocial approach into physiotherapists’ practice. Results The barriers and facilitators to the use of a biopsychosocial approach by physiotherapists were mapped to 10 domains of the TDF (Knowledge; skills; professional role; beliefs about capabilities; beliefs about consequences; intentions; memory, attention and decision processes; environmental context; social influences; emotion). The inclusion of these domains within the TaTT resulted in the identification of 33 BCTs that could foster the use of this approach by physiotherapists. Conclusions Investigating the implementation of a biopsychosocial approach into physiotherapists’ practice from a behavior change perspective provides new strategies that can contribute to successfully implement this approach. Implications for Rehabilitation The implementation of a biopsychosocial approach into physiotherapists’ practice is a complex process which involves behavior changes influenced by several barriers and facilitators. Barriers and facilitators reported by physiotherapists when implementing a biopsychosocial approach can be mapped within 10 domains of the Theoretical Domain Framework. Thirty-three behavior change techniques (e.g., verbal persuasion about capability, problem solving, restructuring the physical environment, etc.) were identified to foster the implementation of a biopsychosocial approach and specifically target barriers and facilitators. By using a behavior change perspective, this study highlights new strategies and avenues that can support current efforts to successfully implement the use of a biopsychosocial approach into physiotherapists’ practice.


Introduction
Like most health professionals, physiotherapists have traditionally received training based on a biomedical model [1]. According to this "structural" model, pain and disabilities are necessarily caused by an injury to a tissue and therefore, the treatment offered should focus on the problematic structures identified [2]. This model has raised several criticisms [3]. Among other things, the biomedical model has been criticized for not taking into account the complexity of pain that we now know is influenced by several psychosocial factors [4]. In addition, the biomedical model does not allow to explain the presence of pain in the absence of structural changes or after tissue healing is expected to have occurred [5]. Furthermore, physiotherapists trained based on a biomedical model notably tend to have maladaptive beliefs, such as high fear avoidance beliefs. They also make recommendations that are not evidence-based [6,7], such as limiting activities and work, as well as encouraging bed rest [8].
Because of these gaps and limitations of the biomedical model, the biopsychosocial model was developed and is now greatly encouraged to support the management of a large variety of people's health conditions, including musculoskeletal [9,10], neurological [11][12][13] and cardiovascular conditions [14,15]. The biopsychosocial model not only considers the physical dimension of a person's experience, but also its psychological and social dimensions [16]. For Borrell-Carrio et al., the biopsychosocial model is not only a philosophy supporting clinical practice, but also a clinical practice approach [17]. Adhering to a biopsychosocial approach encompasses the recognition and understanding of how multiple factors can influence someone's condition, the application of patient-centered care through effective communication, and shared decision making in order to consider patients' individual contexts and meet their needs [18]. Examples of interventions that are coherent with a biopsychosocial understanding of pain include cognitive-behavioral interventions, motivational interviewing, imagery and goal setting [19]. In 2011, Main and George also proposed the use of psychologically informed practice to describe biopsychosocially oriented interventions used by physiotherapists. Psychologically informed practice was presented as the "middle way" between the "traditional" biomedical approach and more cognitive-behavioral approaches [20]. A biopsychosocial approach is therefore a complex clinical practice that can take different forms of interventions and requires several skills on the part of physiotherapists [21].
The use of a biopsychosocial approach is now recommended by several clinical practice guidelines worldwide [10,[22][23][24][25]. It is also strongly recommended for healthcare professionals, including physiotherapists, to use a biopsychosocial approach in providing care for people living with chronic pain [26,27]. Supported by this evidence, a transition towards the use of a biopsychosocial approach by physiotherapists has been observed over the past decade [28][29][30]. However, although the majority of physiotherapists now recognize the importance and relevance of using a biopsychosocial approach in their practice [31], studies show that many physiotherapists continue to adopt an approach based on the biomedical model [32], often reporting lack of skills to adequately address psychosocial factors [31,33].
According to a recent scoping review by Simpson et al. [34] investigating physiotherapists' training to deliver biopsychosocial interventions, most training techniques focused on improving physiotherapists' skills and knowledge. However, changing skills and knowledge may not be sufficient to enact such a change in approach. For example, even if a physiotherapist possesses the knowledge and skills to change a specific behavior, this change in behavior is unlikely to occur if this physiotherapist is unwilling to undertake it or if the environmental context is not favorable. Because of the complex nature of the biopsychosocial approach [35], the uptake of such an approach in the practices of physiotherapists is more likely to occur when combining skill and knowledge training with the integration of behaviour change principles. According to Grol et al., to achieve efficient behavior change, it is crucial to identify the barriers and the facilitators influencing the implementation of the desired behavior and to use specific behavior change techniques (BCTs) targeting those barriers and facilitators [36].
Although studies have explored the barriers and facilitators reported by physiotherapists regarding use of a biopsychosocial approach [37][38][39], no study to date has analyzed these barriers and facilitators using a validated theoretical framework. Using a theoretical framework leads to theoretically informed interventions and generalizable solutions to foster behavior change [40], such as the uptake of a biopsychosocial approach in physiotherapists' practice [41,42]. The Theoretical Domains Framework (TDF) offers a comprehensive framework to allow detailed identification and mapping of the barriers and facilitators encountered by individuals when trying to implement a specific behavior change [43]. Based on the synthesis of 36 behavior change theories and 128 key theoretical constructs, the TDF allows to map barriers and facilitators related to behavior change into 14 domains [44]: 1) Knowledge; 2) Skills; 3) Social/professional role and identity; 4) Beliefs about capabilities; 5) Optimism; 6) Beliefs about consequences; 7) Reinforcement; 8) Intentions; 9) Goals; 10) Memory, attention and decision processes; 11) Environmental context and resources; 12) Social influences; 13) Emotion; and 14) Behavioral regulation. This framework has been used in different health sciences contexts [45][46][47][48][49].
Although mapping the barriers and facilitators within a theoretical framework allows the identification of all domains relevant to a behavior, it provides little information on how best to change this behavior. Recently, researchers from the Human Behavior-Change Project [50] have developed an innovative tool, the Theory and Techniques Tool (TaTT) to identify the most appropriate BCTs aligned with different mechanisms of actions influencing behaviors. Available online (theoryandtechniquetool.humanbehaviourchange.org), the TaTT provides evidence-based links between 74 BCT and 26 mechanisms of action (MoA) influencing a behavior. Interestingly, the MoAs in the tool include the 14 domains of the TDF. Thereby, this tool allows the identification of BCTs that are specific to the TDF domains. Thus, using this tool could allow the identification of BCTs that are directly linked with the barriers and facilitators to the implementation of a biopsychosocial approach reported by physiotherapists.
The use of the TDF together with the TaTT to identify new avenues to support change of practice towards a biopsychosocial approach by physiotherapists has never been done before. A deeper comprehension of the involved barriers and facilitators combined with the identification of specific evidence-based BCTs guided by a theoretical framework could help physiotherapists understand how and where they could act to change their behaviors, and thus foster the implementation of a biopsychosocial approach within their practice.
Hence, the first objective of our study was to map the barriers and facilitators to the implementation of a biopsychosocial approach into physiotherapists' practice within the TDF. The second objective was to identify the specific BCT that could facilitate this implementation using the TaTT.

Materials and methods
This study was carried out in two parts. First, we performed a review of systematic reviews to identify systematic reviews presenting the barriers and facilitators reported by physiotherapists when trying to integrate a biopsychosocial approach within their practice. We used a deductive coding analysis based on the 14 domains of the TDF to map these barriers and facilitators. Second, we used the TaTT to identify BCTs related to the identified TFD domains.

Data search and study selection
With the help of an experienced librarian, a literature search was performed in the Medline, CINHAL, Psychinfo and Embase databases to find systematic reviews addressing barriers and facilitators reported by physiotherapists when trying to implement a biopsychosocial approach into their practice. The search strategies used for each of the databases are available in Supplementary Appendix A. The search was done by the first author on June 15th, 2022. Once duplicates were removed, titles and abstracts of the remaining articles were assessed by the first author (JGH) for their relevance to the subject of the study. Full text reading of the selected articles at this stage was then independently performed by two evaluators to identify the studies to include (JGH, AF). We included systematic reviews (qualitative, quantitative or mixed articles) addressing the barriers and facilitators reported by physiotherapists or their perceptions, behaviors, beliefs or attitudes in regard of the implementation of a biopsychosocial approach within their practice. Systematic reviews addressing the barriers and facilitators of different healthcare professionals were not included if the results of these reviews did not allow the barriers and facilitators specifically encountered by physiotherapists to be isolated. The type of condition treated by physiotherapists was not considered as an inclusion or exclusion criteria. All systematic reviews had to be written in English and peer-reviewed. The two evaluators agreed on all studies to be included.

Assessment of study quality
The second author (AF) assessed the quality of the included studies using the AMSTAR 2 checklist [51]. The AMSTAR 2 is a valid instrument to assess the quality of systematic reviews of randomised or non-randomised studies of healthcare interventions or both [52]. The AMSTAR 2 checklist includes 16 items. Each item is separately reported as "yes", "partial yes" or "no".

Data analysis based on TDF domains
Deductive coding based on the 14 domains of TDF was used to analyse the included systematic reviews and code the barriers and facilitators presented in those reviews [45,53,54]. Names of the domains, their construct definition [44] as well as examples of questions related to each domain [43] were used to inform coding decisions relating to each domain. "Results" and "discussion" sections of each of the included systematic reviews were coded. Citations reported by the authors of the included systematic reviews and from the primary articles were not coded. Since our intention was to identify all the domains that can influence a certain behavior (i.e., TDF domains), no distinction was made between barriers and facilitators. Both were coded into their appropriate domain. The first two authors (JGH, AF) independently coded all included systematic reviews. Coding results were then compared to ensure agreement between both reviewers. When coders agreed, the coding was directly validated. When the coding differed, there was an oral discussion between the reviewers to reach consensus. Reviewers came to an agreement for all codes. For about 10% of the time, reviewers agreed to code content under two domains, since we were interested in identifying all the domains involved, regardless of how often these domains were coded.

Insertion of the identified TDF domains into the TaTT
All identified TDF domains were then inserted into the TaTT via the Theory and Technique Tool website [55]. The TaTT is a table where columns represent 26 MoA, which include the 14 TDF domains, and rows represent 74 BCTs. For the purpose of our study, we only considered the 14 columns (MoA) of the TaTT representing the TDF domains. Cells of the TaTT table (i.e., the junction of a column and a row) represent the strength of an evidence-based link between a BCT (rows) and a MoA (columns). Depending on the strength of a particular link between a MoA and a BCT, the cell on the TaTT table could be white ("absence of evidence"), blue ("non-links"), yellow ("inconclusive") or green ("links") [56]. This strength was determined by a large study by Michie et al. [57] and was established according to three types of studies: a synthesis study of the literature, an expert consensus study and a triangulation study. Since the purpose of our study was to identify specific BCTs shown to be effective in relation to a specific domain, we only considered the green cells ("links") on the TaTT. Thus, if the cell between a column (MoA, in our case, a TDF domain) and a row (BCT) was green, we could assert that the identified BCT was an effective and evidence-based technique to act on the TDF domain involved in the desired behavior change i.e., the use of a biopsychosocial approach by physiotherapists. Figure 1 presents a flow diagram detailing the selection of the included studies. Over five hundred (n ¼ 509) studies were identified and imported from the databases. Once 119 duplicates were eliminated, title and abstract of 390 studies were evaluated. Three-hundred seventy-nine studies were deemed irrelevant to the subject of the study. Eleven studies were therefore included for full text review. This final step excluded seven studies: three studies used an intervention other than the targeted one, three studies targeted a population other than physiotherapists, and one study targeted a different the type of outcome. A total of four studies met the inclusion criteria and were included in this study.

Description of included studies
Characteristics of the included articles are shown in Table 1. All four were systematic reviews published between 2013 and 2020. Two of them reported qualitative articles for which data were collected primarily through semi-structured interviews and focus groups. The other two systematic reviews reported articles with qualitative and quantitative designs, for which data collection methods consisted mainly of semi-structured interviews and surveys. The objectives of the four included systematic reviews were to assess beliefs, perceptions, skills, attitudes, and behavior of physiotherapists towards different aspects associated with a biopsychosocial approach: learning and implementing a biopsychosocial intervention, use of psychological interventions, as well as the assessment and management of cognitive, psychological and social factors. Reviews form Holopainen et al. [58], Synnott et al. musculoskeletal conditions. Whereas the review from Driver et al. [59] included studies looking at physiotherapy in relation to musculoskeletal conditions as well as stroke and schizophrenia. A summary of the quality assessment of included studies using the AMSTAR 2 checklist is available in Appendix B as supplementary material.

Identification of TDF domains
Deductive coding based on the TDF domains of the "results" and "discussion" sections of the included systematic reviews resulted in the mapping of the barriers and facilitators reported by physiotherapists about their use of a biopsychosocial approach within 10 of the 14 domains of the TDF: 1) Knowledge; 2) Skills; 3) Social/professional role and identity; 4) Beliefs about capabilities; 6) Beliefs about consequences; 8) Intentions; 10) Memory, attention and decision processes; 11) Environmental context and resources; 12) Social influences and 13) Emotion.

Domain 1 -knowledge
Analyzed systematic reviews reported that physiotherapists understand the importance and relevance of psychosocial factors in the management of the conditions presented by their patients.
[ … ] there was a general acceptance by the physiotherapists that it was important to assess psychosocial factors [ … ]-Gray et Howe [61] However, physiotherapists seemed to recognize social factors better than psychological factors.
Some physiotherapists seemed to recognise the significant influence on LBP of certain life events, as well as social factors such as the patient's family life and occupational environment. Very little mention of psychological factors was observed. -Synnott et al. [60] Although physiotherapists understood the importance of psychosocial factors, the included reviews showed that they did not always seem to recognize the roles of these factors in contributing to the chronicity of their patient's condition.
In fact, some physiotherapists attributed a progression to chronicity solely to a lack of understanding or awareness of the biomedical and mechanical drivers of pain, with no acknowledgement of the cognitive, psychological and social drivers of chronicity in back pain -Synnott et al. [60] Physiotherapists trained to use a biopsychosocial approach appear to be overwhelmed by the amount of information to assimilate, but some report that training helped them to better understand the involvement of psychosocial factors in the conditions being treated.
[ … ] reported that the biopsychosocial intervention training had changed the way they thought about musculoskeletal pain and its management and reported changing at least some parts of their practice to a more biopsychosocial framework. -Holopainen et al. [58]

Domain 2 -skills
We identified statements discussing the skills physiotherapists need in order to use a biopsychosocial approach. The included reviews mainly depicted the lack of practical skills and the lack of training to improve these skills.
[ … ] current review suggests that psychological skills are not being taught at a level that is making an impact in the physiotherapy setting [ … ] -Driver et al. [59]  Semi-structured interviews (n ¼ 2); group discussion (n ¼ 2); quantitatives questionnaires (n ¼ 6); Delphi conscensus (n ¼ 2 a ); randomized control study (n ¼ 3); analysis of telephone consultations (n ¼ 1) a One study used group discussions and Delphi consensus.
According to the included reviews, while it appeared possible for physiotherapists to receive training on the BPS approach, this training was deemed insufficient

Domain 3 -Social/professional role and identity
The included systematic reviews reported that some physiotherapists perceive the use of a biopsychosocial approach as not being part of their scope of practice. Physiotherapists consider their role to be predominantly focused on a biomedical approach, targeting the mechanical component of pain.
[ … ] this was outside their usual scope of practice [ … ] -Holopainen et al. [58] Many physiotherapists believed that their role was mainly to address the mechanical aspects of LBP [ … ] -Synnott et al. [60] The analyzed systematic reviews also raised several questions about physiotherapists' professional role in the use of a biopsychosocial approach. Some of the included reviews showed that according to some physiotherapists, addressing patient's psychological factors would rather be a role reserved for psychologists.
[ … ] those who were reluctant to assess psychosocial factors expressed that this was due to the facts that either they felt that it was not their professional role to do so [ … ] -Gray et Howe [61] [ … ] many were concerned about the professional role boundaries related to psychosocial factors and about asking questions in psychosocial domains because they considered it outside a physiotherapist's scope of practice [ … ] -Holopainen et al. [58] However, according to the systematic review by Driver et al., some physiotherapists considered providing psychological support to be part of their professional responsibilities.
Psychological support was seen by some PTs in this review as a professional responsibility, and within the boundaries of practice is an essential part of therapy and rehabilitation. -Driver et al. [59] In addition to those questions about the professional role, the use of a biopsychosocial approach appeared to cause confusion between the roles of physiotherapists and psychologists.
The overlapping of roles from physical therapist to psychological therapist may be confusing for both patient and PT [physiotherapist]. -Driver et al. [59] Finally, as reported by Holopainen et al., receiving training related to a biopsychosocial approach would raise critical reflections on the current physiotherapy practice mostly centered on a biomedical model of care.

Domain 4 -beliefs about capabilities
The analyzed systematic reviews reported a lack of confidence by physiotherapists in their ability to use a biopsychosocial approach.
[ … ] they lacked confidence in this area [ … ] -Gray et Howe [61] However, training and experience appear to facilitate their beliefs in their ability to use such an approach.
The physiotherapists reported more confidence in treating people with musculoskeletal pain in a research context and greater confidence in managing musculoskeletal pain in general. -Holopainen et al. [58] Some techniques were used more because the physiotherapists noticed they had good treatment success with them and felt more comfortable using them, or that they were more accepted by patients. -Holopainen et al. [58]

Domain 6 -beliefs about consequences
Beliefs about the consequences of using a biopsychosocial approach on patients' condition and recovery appeared to influence its use. Some physiotherapists tend to underestimate the importance of psychosocial factors. For them, the identification of psychosocial factors seemed to be irrelevant and unnecessary to facilitate their patients' recovery.
However, physiotherapists neither seemed to identify cognitive, psychological or social factors as underlying causes for these observed behaviours, nor considered them as potentially modifiable factors for targeted intervention. -Synnott et al. [60] In contrast, some included reviews showed that physiotherapists trained in a biopsychosocial approach recognized the importance of using such an approach and understood the positive outcome it could have on the care they provide.
[ … ] those who had some level of formal training in psychological interventions held more positive attitudes than those without formal training, and predominantly those with underpinning knowledge valued the importance of psychological skills higher [ … ] -Driver et al. [59]

Domain 8 -intentions
The data analyzed demonstrate that many physiotherapists prefer to use a biomedical and more "mechanical" approach with their patients.
Even among patients who had been told that their LBP was non-specific in nature, physiotherapists preferred to explore the mechanical nature of LBP, either oblivious to the other dimensions of LBP, or choosing not to address it. -Synnott et al. [60] In some cases, some physiotherapists also seem to be unwilling to use a biopsychosocial approach.

Domain 10 -memory, attention and decision processes
Difficulties associated with habit and practice change were also raised. One systematic review mentioned that early implementation of a biopsychosocial approach could foster its use.

Domain 11 -environmental context and resources
Among the included systematic reviews, the main determinants linked to the environmental context were associated to treatment time and the costs associated with longer treatment.

Practice and consultation constraints in the form of time restrictions, cost/ reimbursement issues [ … ] -Driver et al. [59]
For some physiotherapists, one of the biggest barriers to implementing a biopsychosocial approach was that they felt they needed more withinsession time to use the new strategies and techniques effectively. -Holopainen et al. [58] A systematic review also reported that developing skills takes time and practice.
A common theme among many of the studies was that learning and adopting the new role took time and practice. -Holopainen et al. [58] However, some physiotherapists seem to be less affected by the lack of time when they become more confident in using a biopsychosocial approach.
[ … ] some physiotherapists started feeling more comfortable with the time constraints once they gained more confidence in using the new approach and actively tried to find additional time for patients. -Holopainen et al. [58]

Domain 12 -social influences
According to the included reviews, one of the barriers reported by physiotherapists in using a biopsychosocial approach appeared to be the public and patients' perceptions and expectations of physiotherapy.
[ … ] described how patients' biomedical treatment expectations influenced their management approach [ … ] -Synnott et al. [60] According to physiotherapists, the patients would have a biomechanical understanding of pain and would expect to receive treatment based on a biomedical approach.

Domain 13 -emotion
Using a biopsychosocial approach seemed to generate difficult feelings among physiotherapists. For some of them, using a biopsychosocial approach generated fear, frustration and anxiety.
[ … ] felt confused, frustrated and unskilled when it came to treating their clients from a biopsychosocial perspective.
[ … ] -Driver et al. [59] They were afraid of opening "Pandora's box" or a "can of worms" of difficult issues in their patients' lives that they did not feel prepared to deal with [ … ] -Holopainen et al. [58] At the same time, the use of a biomedical approach seemed to bring comfort and happiness.
[ … ] physiotherapists being quite happy to provide advice on local structural diagnoses, and exercise or manual therapies directed at a local mechanical spinal disorder [ … ] -Synnott et al. [60] [ … ] many physiotherapists themselves were more comfortable with LBP presentations that were deemed straightforward and did not involve complicating factors, allowing treatment to focus on 'mechanical' factors such as mobility and movement patterns. -Synnott et al. [60]

Identification of BCTs
The integration within the TaTT of the 10 TDF domains identified through this analysis of systematic reviews made it possible to identify evidence-based BCTs that can act on these specific domains. Table 2 shows the links suggested by the TaTT between TDF  domains (columns -MoAs) and BCTs (rows). Only links (green boxes in the TaTT) are shown and were used in this study. "Biofeedback" BCT is shown in red because even if the TaTT identified there corresponding cell as "links" (green), it was deemed not applicable for this study. Biofeedback refers to the use of an external device that provides feedback on the body. It is therefore difficult to imagine how this technique can be used to change the behavior targeted by this study and among the analyzed population.
As shown in Table 2, the proposed BCTs vary greatly depending on the TDF domains targeted. Across all domains, the TaTT proposed 33 evidence-based BCTs. For example, if a physiotherapist does not believe in his ability to implement a biopsychosocial approach into his practice (TDF domain 4 -Beliefs about capabilities), it would be recommended to use the following BCTs to foster the implementation: problem solving; instruction on how to perform behavior; demonstration of behavior; behavior practice/ rehearsal; verbal persuasion about capability; focus on past success; and self-talk. In another vein, no BCT applying to the "social/ professional role and identity" domain was proposed by the TaTT.

Discussion
Our review of systematic reviews is the first study to use a theoretical framework based on behavior change theories (the TDF) in combination with a tool developed to identify BCTs (the TaTT) to address the implementation of a biopsychosocial approach into physiotherapists' practices. We first used the TDF to map the barriers and facilitators to the implementation of a biopsychosocial approach reported by physiotherapists within the TDF domains. We then used the TaTT to identify BCTs that could best foster the behavior change needed to successfully implement this approach into physiotherapists' practice.
To our knowledge, no study has to date used the TDF to map the barriers and facilitators to the implementation of a biopsychosocial approach into physiotherapists' practice. The use of the TDF highlights how complex behavior change can be, supporting the idea that implementing a biopsychosocial approach is much more than just using a specific intervention [18] and that "shifting practice to a biopsychosocial approach requires a paradigm shift in beliefs and behavioural change" [34]. Other studies have also found that a change of practice, for example implementing practical clinical guidelines, implies many domains of the TDF [53,[62][63][64]. Depending on the practice change targeted by these studies, three to 12 TDF domains were identified. These elements confirm the great complexity of changing professionals' clinical practices and the importance of using adequate strategies to foster such changes.
A recent systematic review of qualitative articles by Ng et al. [65] synthesized the evidence concerning the barriers and facilitators to the use of a biopsychosocial approach by health professionals (acupuncturists, doctors, chiropractors, nurses, occupational therapists, physiotherapists, osteopaths, pain management consultants and psychologists) to manage musculoskeletal pain. This study was not included in our review since the results did not allow to isolate the barriers and facilitators specifically encountered by physiotherapists. Furthermore, it did not map the barriers and facilitators within a theoretical framework. The authors identified several barriers and facilitators that were similar to those identified in our review: lack of knowledge and skills in dealing with psychosocial factors; patients' expectations and beliefs towards more biomedically oriented interventions; professional roles; as well as emotions such as fear and frustration. This systematic review also demonstrated that those barriers and facilitators originated from other sources than the individuals (i.e., health professionals) and rather extended to patients (e.g., lack of motivation and seeking passive treatment); context of care (e.g., access to other health professionals); reimbursement and funding (e.g., insurance patient health), organizational practices (e.g., consultation time and waiting lists); training programs (e.g., little training in communication skills); and governmental policies (e.g., influences of the role and funding of governments to remunerate health professionals). Although our study made it possible to identify the barriers and facilitators perceived by physiotherapists, Ng's review demonstrated that it remains important to consider the whole range of factors that could act as barriers and facilitators to the implementation of a biopsychosocial approach.
Our findings also permitted the identification of specific BCTs that could help to successfully implement a biopsychosocial approach into physiotherapists' practice. This is different than previous studies that focused on providing techniques to enhance physiotherapists' skills and knowledge about this approach. According to Demmelmaier et al.'s study, interventions aimed at changing complex behaviors are unlikely to be effective if they focus only on the knowledge and skills associated with a behavior [66]. Our findings also align with their conclusions that BCT should be tailored to the barriers perceived by individuals who want to change this behavior and include interventions such as personal goal setting, skills training, and ongoing performance feedback. Some of these strategies were used in a 2019 mixedmethods study exploring the dose, reach and participation of physiotherapists using implementation interventions based on behavior change. This study was underpinned by social cognitive theory to implement a biopsychosocial approach in physiotherapists' practice [67]. In their study, Fritz and colleagues mentioned that outreach visits, peer coaching, educational material and individual goal-setting were useful techniques to facilitate behavior change. They also highlighted the complexity and diversity of mechanisms involved during a clinical behavior change process.
As introduced previously in a recent scoping review, Simpson et al. [34] explored the wide range of techniques used to train physiotherapists to foster their use of biopsychosocial interventions to treat musculoskeletal pain conditions. The authors recognized that the training currently offered to physiotherapists varied enormously. Some training techniques were short and simple, mainly through didactic education such as lectures, while others were longer and more sophisticated, involving workshops, supervision, and role play. Although the authors could hardly conclude on the effectiveness of some of these techniques, most of them aimed at improving physiotherapists' knowledge and skills and not at changing their behavior. This difference is crucial. According to our findings, physiotherapists effectively raised barriers or facilitators regarding their knowledge and skills, but these elements only represent two of the 10 TDF domains identified. Consequently, the recent emphasis on targeting physiotherapists' skills and knowledge to foster the implementation of a biopsychosocial approach could be a reason why it has been difficult to successfully implement this approach. By addressing the implementation of a biopsychosocial approach from a behavior change perspective, our study considers the wide spectrum of barriers and facilitators involved and proposes novel strategies to specifically target the different barriers and facilitators involved.

Strengths and limitations
This is the first review of systematic reviews using the Theoretical Domains Framework (TDF) and Theory and Technique Tool (TaTT) to address the barriers and facilitators reported by physiotherapists to implement a biopsychosocial approach into their practices in combination with the techniques that could foster this implementation. The use of a framework and a tool derived from behavior change theories brings valuable knowledge to better understand the extent of factors involved and to guide the interventions towards an efficient change of practice.
As for limitations, since data coding was not done on primary data (e.g., interview transcripts), information surrounding the context of statements was sometimes missing. Having more contextual information could have led to a different coding. Moreover, since we chose to code data from systematic reviews and not from primary studies, it was not possible to calculate the weight or frequency of every coded domain compared to one another. Doing so by using a different methodology could have shown which domains (i.e., barriers and facilitators) are more or less frequently encountered by physiotherapists. Occasionally, some content was coded under two TDF domains. We chose to code some content under two domains to avoid arbitrary choices, and to ensure that we covered all possible domains, regardless of the weight of each of these domains. For example, one of the included studies stated that when using a biopsychosocial approach, physiotherapists were afraid of opening a "pandora's box" because they thought they would not know how to deal with the issues raised by the patients. In this situation, we decided to code the citation under domains 13, emotions, as well as domain 4, belief about capabilities. In our view, this shows the complexity of the multiple domains influencing the use of a biopsychosocial approach by physiotherapists.
Also, although our search strategy didn't specifically target studies looking at physiotherapy related to specific conditions, most of the included reviews in our review addressed the barriers and facilitators encountered by physiotherapists managing musculoskeletal conditions. Therefore, it seems that past studies have mainly been interested about the barriers and facilitators encountered by physiotherapists working with people having musculoskeletal conditions. Despite this tendency oriented towards a specific population, it is possible that the results of our study could also be relevant to physiotherapists working with people presenting various conditions as most barriers and facilitators reported in our study do not seem specific to musculoskeletal conditions. Moreover, as mentioned in the discussion, the barriers and facilitators to the implementation of a biopsychosocial approach can arise from sources other than individuals, such as organizations for example. Although the TDF was developed from behavioral theories of individuals as well as organizations, few TDF domains relate to organizational characteristics [45,46]. To reduce this limitation, it could have been relevant in our analysis to combine the use of the TDF with another theoretical framework. For example, a combined use of the TDF and the Consolidated Framework for Implementation Research (CFIR), could have provided a wider understanding of the barriers and facilitators involved, by focusing more on potential organizational influences [68].
Finally, we decided to consider and present only the BCTs from the green boxes ("links") of the TaTT, regardless of the white ("absence of evidence"), blue ("non-links") and yellow ("inconclusive") boxes. We made this choice based on our intention to link the identified TDF domains to evidence-based BCTs (green boxes). However, considering yellow boxes could offer more information on possible BCTs since "inconclusive" evidence does not rule out a possible link between a BCT and a TDF domain. Moreover, the list of BCTs identified in our study by the TaTT is not intended to be a rigid and exhaustive list as the TaTT was developed to be an interactive tool always evolving according to scientific evidence.

Conclusion
In sum, investigating the barriers and facilitators to the implementation of a biopsychosocial approach into physiotherapists' practice from a behavior change perspective highlights the complexity of this implementation process. Our study mapped the barriers and facilitators reported by physiotherapists into multiple TDF domains. For each of them, BCTs specifically related to those domains were identified with the TaTT. These data provide new strategies that can contribute to novel initiatives leading to successful implementation of a biopsychosocial approach into physiotherapists' practices. A better tailored approach, aimed to change behavior, could be a crucial key in the integration of a BPS approach in physiotherapists' practices. Future research assessing the effectiveness of the specific BCTs carefully chosen and linked to specific domains of behavior change identified in this review could help reach important progress in this field. Such studies could help researchers and clinicians better understand how targeting domains of behavior change can influence the choice, acceptability and effectiveness of specific BCTs to implement a biopsychosocial approach with success.