Introduction
The development of empirically grounded clinical interventions (EGCIs) typically follows an iterative process that draws on theory, clinical practice, outcome research and experimental studies, and applies the scientist-practitioner approach (Salkovskis, Reference Salkovskis2002). This has been used to good effect in developing psychological interventions for a variety of mental health conditions such as panic disorder, social anxiety disorder and post-traumatic stress disorder (Clark, Reference Clark2004; Salkovskis et al., Reference Salkovskis, Sighvatsson and Sigurdsson2023), but has been less commonly applied to the processes of therapy.
One such process is that of engagement in therapy. Engagement is thought to be a critical component of cognitive behavioural therapy (CBT) and other psychological therapies (Glenn et al., Reference Glenn, Golinelli, Rose, Roy-Byrne, Stein, Sullivan, Bystritksy, Sherbourne and Craske2013). You can develop the most effective treatment in the world, but it is guaranteed not to work if the patient is not there, or if they do not understand, remember, or agree with what is being discussed. There is good evidence to suggest that signs of engagement such as attending more sessions (Clark et al., Reference Clark, Canvin, Green, Layard, Pilling and Janecka2018) and completing more and higher quality homework (Kazantzis et al., Reference Kazantzis, Whittington, Zelencich, Kyrios, Norton and Hofmann2016; Kazantzis et al., Reference Kazantzis, Brownfield, Mosely, Usatoff and Flighty2017; Mausbach et al., Reference Mausbach, Moore, Roesch, Cardenas and Patterson2010) are reliably associated with better outcomes in CBT. However, given this potential importance it is perhaps surprising that engagement has been less subject to systematic empirical investigation, and more commonly measured using limited behavioural indicators or discussed in terms of the therapist’s felt sense.
When we examine the literature we see that when therapy engagement is measured in empirical studies, it is done in numerous and inconsistent ways, highlighting a lack of consensus in how it is defined and conceptualised (Holdsworth et al., Reference Holdsworth, Bowen, Brown and Howat2014; Tetley et al., Reference Tetley, Jinks, Huband and Howells2011). Measures include treatment uptake, treatment expectancy, treatment readiness, motivation, session attendance, session participation, time spent on homework, homework compliance, in-session involvement, therapeutic alliance, therapist ratings of commitment, understanding of CBT principles, drop-out rates, treatment interruptions and various others (Albaum et al., Reference Albaum, Sellitto, Vashi, Bohr and Weiss2024; Biswal et al., Reference Biswal, Gandhi, Singla, Velleman, Zhou, Fernandes, Patel, Prina, Sequeira, Garg, Bhatia and Nadkarni2024; Bourke et al., Reference Bourke, Barker and Fornells-Ambrojo2021; Chavira et al., Reference Chavira, Golinelli, Sherbourne, Stein, Sullivan, Bystritsky, Rose, Lang, Campbell-Sills, Welch, Bumgardner, Glenn, Barrios, Roy-Byrne and Craske2014; Glenn et al., Reference Glenn, Golinelli, Rose, Roy-Byrne, Stein, Sullivan, Bystritksy, Sherbourne and Craske2013; Kerns et al., Reference Kerns, Burns, Shulman, Jensen, Nielson, Czlapinski, Dallas, Chatkoff, Sellinger, Heapy and Rosenberger2014; LeBeau et al., Reference LeBeau, Davies, Culver and Craske2013).
Metrics such as session attendance or homework completion have been described as ‘behavioural manifestations of engagement’ (Chavira et al., Reference Chavira, Golinelli, Sherbourne, Stein, Sullivan, Bystritsky, Rose, Lang, Campbell-Sills, Welch, Bumgardner, Glenn, Barrios, Roy-Byrne and Craske2014) and are quite commonly used in studies given they are easy to define and measure. However, I suggest that relying solely on such behavioural measures fails to capture the richness and complexity of the engagement concept. Consider a patient who is only completing homework tasks to please their therapist, but who does not see the point of homework or the need to be doing anything between sessions. Or a client who does not want therapy but is only attending sessions because their partner has said they will leave them if they do not sort out their problems. Both clients would be considered ‘engaged’ based on behavioural indicators alone, when clearly there are other engagement issues present. Taking a broader focus may be helpful to provide a more comprehensive conceptualisation of what appears to be a multi-faceted construct. Below, I offer a conceptual model grounded in clinical experience that aims to capture the many elements that represent or contribute to an overall sense of engagement in therapy.
Model description and discussion
As shown in Fig. 1, the model divides engagement into the three facets described below. The arrows highlight that the facets are not independent and are likely to influence each other. Although this model was developed with CBT in mind, most elements within it are not necessarily CBT-specific, meaning it may be of use and relevance across other therapy modalities.
A conceptual model of engagement in psychological therapy (adapted from Thew, Reference Thew and RyumIn press).
Engagement with therapist
A core component of engaging with a therapeutic intervention is for the patient to feel able to work together meaningfully with the therapist. This facet therefore encompasses established concepts such as the therapeutic alliance (Elvins and Green, Reference Elvins and Green2008; Hougaard, Reference Hougaard1994) which shows a reliable relationship with clinical outcomes (Bourke et al., Reference Bourke, Barker and Fornells-Ambrojo2021; Cameron et al., Reference Cameron, Rodgers and Dagnan2018; Flückiger et al., Reference Flückiger, Del Re, Wampold and Horvath2018), although not necessarily through a direct causal mechanism (Salkovskis et al., Reference Salkovskis, Sighvatsson and Sigurdsson2023). It also includes more basic processes within therapy sessions such as the extent to which the patient is willing and able to communicate and interact with their therapist, how involved they feel in sessions, and the amount of collaboration present when discussing and planning therapy activities. These basic processes are likely necessary to establish a strong alliance, which may in turn promote further engagement behaviours (e.g. regular session attendance, homework completion).
Ideally, we would hope patients feel able to ‘bring their whole selves’ to therapy and feel comfortable to talk about topics that might otherwise be avoided due to discomfort, embarrassment or shame. Patients’ past experiences, cultural and/or religious backgrounds and other demographic and clinical characteristics will affect how they relate to therapists (and vice versa) and could therefore facilitate or pose a barrier to effective engagement.
Engagement with therapy approach
This facet includes a range of elements that relate to patients’ perceptions of therapy. This covers the extent of patients’ understanding of, and agreement with, the overall idea of therapy, the specific therapy approach being used (e.g. CBT), the format of treatment (e.g. in-person, remote delivery, internet/app based), and individual therapy techniques (e.g. behavioural experiments, imagery rescripting, etc.). Reservations, misunderstandings or concerns in this area (e.g. ‘I’m not sure this therapy is right for me’; ‘I’d worry about what to write in a therapy app’) are likely to lead to difficulties in engagement so are important to identify and address early where possible.
Established concepts in the literature linked to this facet include perceived treatment credibility (Constantino et al., Reference Constantino, Coyne, Boswell, Iles and Vîslă2018), e.g. beliefs about whether the treatment seems logical and is likely to be successful, and ‘socialisation to the model’, an idea that recognises that patients cannot be expected to know how therapy and specific approaches work, so may need guidance and support to understand and develop hope around this (Daniels and Wearden, Reference Daniels and Wearden2011; Mahoney-Davies et al., Reference Mahoney-Davies, Roberts-Collins, Russell and Loades2017; Roos and Wearden, Reference Roos and Wearden2009). In broader therapy terms, this facet also includes concepts such as readiness for change (Crane et al., Reference Crane, Hotton, Shelemy and Knowles-Bevis2024), motivation for treatment (Cahill et al., Reference Cahill, Adinoff, Hosig, Muller and Pulliam2003; Simoneau and Bergeron, Reference Simoneau and Bergeron2003), and self-efficacy (Bandura, Reference Bandura1977). Effective engagement in therapy is likely to be much more possible when a patient feels it is the right time, they feel motivated to understand and address their difficulties, and feel they have the skills and abilities to do so. Of course, these may not all be in place at the outset of treatment and may need to be explored and developed by the therapist.
Practical engagement
This facet includes the ‘behavioural manifestations’ of engagement (Chavira et al., Reference Chavira, Golinelli, Sherbourne, Stein, Sullivan, Bystritsky, Rose, Lang, Campbell-Sills, Welch, Bumgardner, Glenn, Barrios, Roy-Byrne and Craske2014) that are more commonly used as indicators of engagement in the literature. It includes attending therapy sessions (or equivalents in other therapy formats such as taking telephone calls, responding to messages, or working on an online platform), punctuality, and completing the ‘tasks’ of therapy, such as homework assignments, questionnaires, and/or preparing for the next session. Many of these elements stem from concepts originally termed treatment ‘compliance’ (Kazantzis et al., Reference Kazantzis, Deane and Ronan2004; LeBeau et al., Reference LeBeau, Davies, Culver and Craske2013) and later ‘adherence’ (Biswal et al., Reference Biswal, Gandhi, Singla, Velleman, Zhou, Fernandes, Patel, Prina, Sequeira, Garg, Bhatia and Nadkarni2024). As suggested above, these elements are generally easy to measure and clearly demonstrate aspects of engagement, but may not provide the full engagement picture until considered in conjunction with the other model facets.
Aims and potential uses of the model
By taking a broad perspective and bringing together a range of constructs related to engagement, it is hoped this model provides a comprehensive way to conceptualise engagement that may be of use to clinicians and researchers. In clinical practice it may offer a simple and structured way to review patients’ engagement, conceptualise difficulties, and suggest routes forward. For example, a patient might be interacting and getting on well with their therapist (stronger engagement with the therapist) but appears ambivalent about therapy tasks (weaker engagement with therapy approach) and is not doing much between sessions (weaker practical engagement). In this case, it may help to draw on the strength of the relationship to discuss any concerns or confusion about the treatment approach, and any barriers to between-session work they are experiencing (for further examples see Thew, Reference Thew and RyumIn press).
It might therefore be plausible to view engagement difficulties and therapeutic ruptures as a mismatch between levels of engagement across the three facets. Given that each facet may change over time and influence the others, the model aligns with other theoretical accounts that emphasise the dynamic and interactional nature of the alliance (Safran and Muran, Reference Safran and Muran2000). Aiming to achieve strong and consistent engagement that is balanced between the facets may therefore provide the optimal conditions for supporting clinical progress and recovery. Examining the components of recovery frameworks such as CHIME (Connectedness, Hope, Identity, Meaning, Empowerment) (Leamy et al., Reference Leamy, Bird, Boutillier, Williams and Slade2011), we can see how these are most likely to be fostered during therapy when engagement within and across the three facets is strong and balanced.
It is also hoped that the model may be of use by offering directions for further research. This may involve developing more comprehensive tools to measure engagement rather than relying on single indicators such as session attendance. The model itself would benefit from empirical examination, which could be done in various ways. Firstly, it would be helpful to explore in greater depth the relationships between the elements of each facet. For example, patient-rated involvement in sessions, clinician ratings of in-session collaboration, and observer-rated interaction in sessions may all represent distinct constructs, but could also strongly overlap with each other or with therapeutic alliance.
Secondly, the proposed interactions between the facets should be tested. For example, there is evidence indicating that elements of the ‘Engagement with therapy approach’ facet, such as holding positive beliefs about treatment, and understanding the rationale for CBT, are reliable predictors of between-session activity (Bennion et al., Reference Bennion, Lovell, Blakemore, Vicary and Bee2024). It would be helpful to explore further the reverse relationship (whether practical engagement supports greater engagement with the therapy approach) and the other between-facet relationships.
Thirdly, it would be important to analyse the facets’ predictive associations with symptom change and overall clinical outcomes (both independently and in combination). More broadly, it would also be valuable to develop a greater understanding of patients’ experiences across the three facets, as a first step to exploring ways to support greater engagement in each. Relatedly, further examination of engagement difficulties, disengagement, and drop-out could provide important insight into how we can best support patients to get the most from treatment.
Further research directions
In addition to a need for studies on engagement in general, there are three specific areas where I wish to highlight potential gaps in our understanding that would benefit from further research attention. First, I would suggest that at present, there can be a tendency for patients’ engagement in CBT to be determined and characterised primarily by the therapist (e.g. ‘Patient X is showing limited engagement in treatment’), with less attention paid to patients’ own perspectives. This risks the assumption that engagement difficulties are the patient’s problem, when in reality the problem may be located with the therapist, the therapeutic relationship, the intervention, or a combination of these. In many cases, a patient ‘failing to engage’ might be more accurately characterised as the therapist failing to engage the patient. We need to understand more about the factors that might help or hinder each facet of the engagement process, and the strategies that might best foster meaningful engagement. This would benefit from further systematic enquiry into patients’ therapy experiences (e.g. McPherson et al., Reference McPherson, Wicks and Tercelli2020) so these voices can be heard and engagement topics can be explored directly.
Second, we are now seeing a greater and long overdue focus on researching therapy delivery across a wider range of countries and cultural contexts, and the therapy experiences of people from minoritised groups (Alam et al., Reference Alam, O’Halloran and Fowke2024; Khairat et al., Reference Khairat, Hodge and Duxbury2023; Naeem et al., Reference Naeem, Phiri, Nasar, Gerada, Munshi, Ayub and Rathod2016). There is evidence highlighting that patients’ cultural or religious backgrounds may have a significant bearing on their expectations and beliefs about whether and how therapy will work and how successfully they could work with a therapist who might be from a different cultural background (Hassan et al., Reference Hassan, Lack, Salkovskis and Thew2024; Hassan et al., Reference Hassan, Lack, Salkovskis and Thew2025). This has the potential to pose engagement challenges, particularly engagement with the therapist in the critical early phase of treatment. Further research to explore the interaction between cultural context and engagement is much needed. This would also include topics such as the cultural adaptation of treatments (Bernal et al., Reference Bernal, Jiménez-Chafey and Domenech Rodríguez2009; Naeem et al., Reference Naeem, Phiri, Nasar, Gerada, Munshi, Ayub and Rathod2016) to examine the extent to which, and potential mechanisms of how culturally relevant treatment content might promote greater engagement with therapy materials and subsequently improve clinical outcomes.
Third, now that therapy interventions are delivered using a range of different formats and digital technologies, traditional understandings of what it means to engage in therapy may need further elaboration. One interesting example of this is the idea that in internet interventions, it has been highlighted that patients may develop an alliance with both the therapist and the internet platform itself (Kiluk et al., Reference Kiluk, Serafini, Frankforter, Nich and Carroll2014; Ormrod et al., Reference Ormrod, Kennedy, Scott and Cavanagh2010). This could lead to more complex patterns of engagement than might be encountered in ‘traditional’ in-person therapy. For example, in a therapist-guided internet intervention, a patient may not be answering calls from their therapist (typically a sign of disengagement), but perhaps they have been logging into the treatment website and completing treatment modules (a sign of engagement). Therefore, knowing their engagement with the internet platform provides additional context and allows us to formulate the disengagement differently (e.g. perhaps they are finding the treatment helpful but are feeling anxious about calls with their therapist). Treatment format needs to be considered within the ‘Engagement with therapy approach’ facet because advances in therapy delivery may require more nuanced understandings of how patients interact with and engage in treatment.
Conclusion
The literature suggests that engagement is an important factor in successful therapy, but that it is defined and conceptualised inconsistently, with few studies examining engagement processes specifically. Taking the empirically grounded clinical interventions approach and scientist-practitioner perspective may be a productive route forward. The model presented aims to start this process by bringing together existing theoretical ideas and observations from clinical practice, offering a framework to support further research in this area. Paying greater attention to the process of engagement, both at the level of individual clinical practice and larger scale studies, will likely prove valuable in helping people to get more from therapy sessions, reducing drop-out, and enhancing clinical outcomes.
Data availability statement
Data availability is not applicable to this article as no new data were created or analysed in this study.
Acknowledgements
Thanks to Dr Aimee McKinnon for comments on an earlier version of this article.
Author contributions
Graham R. Thew: Conceptualization (lead), Writing - original draft (lead), Writing - review & editing (lead).
Financial support
This work was supported by the Medical Research Council [grant number MR/Z504087/1].
Competing interests
The author declares none.
Ethical standards
The author has abided by the Ethical Principles of Psychologists and Code of Conduct as set out by the BABCP and BPS. This work did not recruit participants or examine existing data so participant consent and ethical review procedures were not applicable.
Open access
Comments
No Comments have been published for this article.