Background

Dialectical Behaviour Therapy (DBT) [1] synthesises behavioural based therapy components (orientated towards change) with elements from mindfulness (orientated towards increasing acceptance). DBT is typically offered to people with a diagnosis of Borderline Personality Disorder (BPD) and a history of suicidal and self-harming behaviour. Therapists aim to impart new skills and develop clients’ behavioural flexibility to draw on appropriate skills in any given social or emotional situation. Core treatment components include individual therapy, telephone skills coaching, skills group and a clinician consultation team (where DBT therapists access support and guidance from other DBT team members), although services may only deliver some of these components (e.g. [2]). Several reviews summarising the evidence for DBT effectiveness are available (e.g. [3, 4]).

Implementation is the process through which the uptake of evidence-based interventions in routine clinical practice is systematically promoted. Transdisciplinary implementation frameworks exist, for instance, Promoting Action on Research Implementation in Health Services (PARIHS: [5, 6]), the Consolidated Framework for implementation for implementation research (CFIR: [7]), and the Core Implementation Components model ([8, 9]). The PARIHS framework covers many of the core elements of these models: context, evidence, facilitation and intervention elements. Context refers to the environment or setting that the implementation takes place in. Evidence can be derived from research, clinical experience or patient preference. Facilitation refers to the people and processes that support implementation and the intervention element demotes the characteristics of the intervention to be implemented. CFIR has an additional element related to implementation processes, which describes the practical implementation tasks undertaken. Each of these elements are sub-divided. For instance, PARIHS sub-divides evidence into research and published guidance, clinical experience and professional knowledge, preferences and experiences, and local knowledge. National implementations refer to many of these elements in their guidance, for example, the Increasing Access to Psychological Therapies manual [10].

DBT has unique features, such as, a multicomponent therapy process, telephone skills coaching, and a consultation team. The characteristics of people with BPD (the core client group) could also necessitate bespoke implementation strategies. The question of how best to implement a DBT intervention arose in the context of the Enabling and Motivating people (with a Personality Disorder) in Occupation, Wellbeing, Education and Responsibility (EMPOWER) research programme (NIHR Programme Grant: RP-PG-1212-20,011), which is developing and evaluating a DBT- Skills for Employment (DBT-SE) intervention. The evidence about DBT implementation has seldom been reviewed and this work was undertaken with a view to developing an implementation toolkit for the DBT-SE intervention. We aimed to review the DBT implementation literature to develop and refine a bespoke DBT implementation framework.

Methods

Framework development

To create an initial DBT implementation framework, elements from the main transdisciplinary frameworks (PARIHS, CFIR, and the Core Implementation Components model) and from published DBT implementation guidance [11, 12] were synthesised. PARIHS [5, 6] was selected as the underpinning framework as it highlights the pivotal role of contextual factors. Although it developed from existing models, this is the first framework to incorporate implementation insights from the DBT literature. The initial framework is represented in Fig. 1 and is described in Additional file 1.

Fig. 1
figure 1

DBT implementation framework: first iteration. Key: PARIHS, 20041; Damschroder et al., 20092; Fixsen & Blasé, 20093, Swales 2010a, 2010b4

Full size image

Critical literature review

To refine the framework, a critical literature review [13] was conducted. In critical reviews, the synthesis process is used to create a new model or a model embodying existing theory which then provides a ‘launch pad’ for subsequent testing. One of the strengths of this type of review lies in the analysis undertaken to create the model [13].

Four databases were searched with the terms ‘DBT’ and ‘Implementation’ in January 2016; Medline (EBSCO), CINAHL (EBSCO), PsycInfo (ProQuest), and PubMed (NCBI). These databases were selected as they hold health and psychology related literature. Reference lists of included papers were additionally screened as implementation issues might be discussed without this term being used as a key word or included in the abstract or title. DBT was defined as any combination of components or interventions which were identified as DBT by the study authors (Additional file 2 contains an example search). Implementation was defined as the process of introducing and sustaining DBT in routine practice. All retrievals were managed in RefWorks, an online bibliographic management programme. Only peer-reviewed papers were included but no date or evidence type restrictions were applied. For resource and time reasons, only papers published in English were included. The first author conducted the review and the eligibility of database retrieved papers was checked by a second reviewer (reviewer agreement was 97% with all disagreements resolved through discussion).

Consistent with the critical review approach, papers were not excluded for methodological reasons [13]. However, prospective and retrospective studies of implementation were considered to provide the strongest evidence due to their explicit focus on implementation. Discussion pieces were judged to form the weakest evidence as the experiences they are based on are often not accessible for review. The data extracted from papers included; the design, context, methodology, implementation barriers and facilitators, as well as author conclusions and recommendations (Additional file 3 contains the data extraction template). Extracted data were discussed by the review team and where necessary the text was re-reviewed.

Evidence synthesis

Extracted data relating to implementation barriers and facilitators, conclusions and recommendations were coded using deductive content analysis. This process used the elements and sub-elements in the DBT implementation framework as code labels and assigned them to the data segments. Where the extracted data did not fit any existing codes, a new code name was added and this process continued until all the data were categorised. Coding was conducted by the first author and a second reviewer checked the coding applied to sixteen papers (10 % of the papers coded): although conservatively judged agreement was 66% (a criterion that the same sub-elements were coded in each paper), differences in coding were negligible and easily resolved through discussion. For instance, the most common cause for disagreement was which code best captured the data. The team reviewed the final synthesis to ensure it presented an accurate reflection of the data.

Results

Critical literature review

Sixty-two papers met the inclusion criteria (32 from database and 30 from reference list searches), although two papers were unobtainable within the time limit of the review. The main reasons for exclusion were failure to consider DBT or a failure to discuss implementation issues (see Fig. 2). There were 11 discussion papers ([11, 12, 14,15,16,17,18,19,20,21,22]) and as these were considered the weakest form of evidence they were not included in the synthesis, but are detailed in Table 1. As seen in Table 2, nine papers collected retrospective ([23,24,25,26,27,28,29,30,31]) and five papers ([32,33,34,35,36]) collected prospective implementation data. There were 16 programme descriptions ([37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52]) and 19 trial process analyses ([53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71]). The majority (N = 38) of papers were from the United States (US), and most implementations of DBT were in mental health services.

Fig. 2
figure 2

Literature review flow chart

Full size image
Table 1 Discussion papers
Full size table
Table 2 Implementation papers, programme descriptions and trial process analysis papers
Full size table

Evidence synthesis

Overall, 788 framework codes were assigned to the extracted data: 170 codes were allocated to studies specifically considering implementation, 209 codes to process analysis studies, 224 codes to programme descriptions, and the remainder were assigned to discussion papers (see Additional file 4). The DBT implementation framework is used to present the literature synthesis and, when possible, the data discussed is derived from the papers which explicitly studied implementation.

Context

Our initial DBT framework included seven context sub-elements (culture, leadership, evaluation, goal fit and suitability, individual characteristics, facilitative administrative supports and system interventions). Our synthesis of the literature yielded five primary sub-elements (culture, leadership, goal fit and suitability, facilitative administrative supports and system interventions), which are discussed below:

Culture

There are two elements of culture that capture staff behaviour within the organisation [5, 6]:

Communication processes

Better ratings of organisation cohesion and communication correlated with the implementation of more DBT components [26], perhaps because institutional adoption of DBT depends on the collaboration of many staff [34]. On-going external consultation helps achieve sustainable programmes [23] and good communication was important within the DBT consultation team [29]. There were examples of communication forming both a facilitator (e.g. [40]) and a barrier [46]. Communication within and across teams seemed particularly important when client characteristics, such as intellectual disability or offender status, meant collaborative working was essential (e.g. [45]).

Climate

Higher scores on the Team Climate Inventory correlated with the implementation of more DBT components [26] and limited understanding of staff and patients’ needs could form a barrier [29]. The importance of team support was endorsed by therapists [28]. Attitudes toward BPD seemed key. A non-judgemental, validating stance seems necessary to create the right environment [25], and better attitudes towards BPD correlated with increased use of DBT [35]. In one survey negative administrator attitudes reportedly impeded implementation [23].

Leadership

In a therapist survey, one of the most common reasons for DBT team cessation was a lack of leadership or organisation ‘buy-in’. Where team leadership was supportive, 19.6% of respondents said this facilitated implementation [31]. Respondents in another study similarly reported that a lack of understanding amongst service leaders constituted an implementation barrier [28]. Often the implementation of DBT had not been pre-planned and in these scenarios having a ‘DBT champion’ in the organisation seemed important. Champions needed to have influencing skills (e.g. [51]), cultural sensitivity, a willingness to undertake tasks, such as, securing funding [43], and an ability to model DBT skills [46]. In many cases the DBT consultation team seemed to undertake championing tasks through generating interest (e.g. [59]), establishing collaborations (e.g. [65]), offering expertise to other agencies (e.g. [52]) and providing support to the wider staff team (e.g. [69]).

Goal fit and suitability

Sometimes DBT was viewed positively from the outset [36] and greater confidence in DBT effectiveness correlated with increased use of DBT [35]. However, DBT was not always seen as suitable [24, 30]. DBT implementation was also weakened by competing service priorities [31]. For instance, in a substance abuse service, DBT was incompatible with the delivery model of short visits primarily providing methadone [23]. Some administrators were concerned about the telephone coaching component of DBT, as telephone support had not worked previously [36] and services need a minimum number of patients to run DBT groups [30]. However, whilst belief in DBT suitability and fit could facilitate implementation (e.g. [58]), the lack of this belief was not necessarily a barrier, as perceptions could change during the implementation process (e.g. [55]).

Facilitative administrative supports

Insufficient time could be a barrier, whereas the allocation of sufficient time could be a facilitator [31]: in one survey, 42% of therapists reported having a lack of time to provide DBT [23]. Some therapists talked about needing to divide their time between different tasks [28] and administrators were concerned that DBT training would keep staff from their clinical duties [36]. Other required resources were finances [36] and space: having adequate space correlated with the implementation of more DBT components [26]. The data also suggests that contingency management has the potential to influence implementation. For instance, organisations often failed to reduce other staff-held responsibilities to compensate for new DBT commitments [23] thereby punishing engagement in DBT. Enabling natural contingencies, such as, smaller caseloads and enabling staff to hold a highly visible role in the service seemed more effective (e.g. [40]) than providing tangible reinforcements (e.g. [65]) - although see [48] for an exception.

System interventions

In the US services need to ensure they receive sufficient referrals to remain viable and so coordination with external agencies is necessary [36]. There were five examples of coalitions facilitating implementation ([14, 42, 54, 55, 59]). One research group suggested that training courses and merging consultation teams might foster coalitions [23] and there was an example of a service establishing two consultation teams: one service-led, the other interagency [45].

Evidence

Informed by PARIHS our initial framework referred to the sub-elements of research and published guidance, clinical experience and professional knowledge, preferences and experiences and local knowledge. However, our search yielded just two primary sub-elements (preferences and experiences and local knowledge and evaluation):

Preferences and experiences

Some therapists expressed a preference for DBT [28] but 47% of therapists said there were challenges in recruiting sufficient patients [23]. Patients reported that they liked many aspects of DBT [29], though they need sufficient cognitive capacity to understand DBT skills and this may constitute a barrier for some [25]. The literature contained evidence that recruitment (e.g. [38]) and attrition (e.g. [52]) could be a problem and there were many attempts to reduce attrition including: ensuring participation was voluntary (e.g. [51]), careful selection of patients (e.g. [42]), providing more information about what DBT would entail (e.g. [53]) and, when appropriate, involving caregivers (e.g. [46]). On two occasions tangible reinforcement was offered [40, 64].

Local knowledge and evaluation

Evidence of clinical improvement can reinforce implementation attempts [24], although only 7% of respondents in one survey agreed that improved patient outcomes were an implementation facilitator [31]. Sometimes demonstrating good patient outcomes generated interest in DBT [37] and led to further funding [43]. However, there were only five examples of services routinely evaluating outcomes ([37, 39, 42, 47, 48]).

Facilitation

Our initial framework referenced six sub-elements (strategies, support, training, coaching and ongoing consultation, facilitator skills/ qualities and recruitment and selection). However, our search and synthesis yielded two primary sub-elements (team capacity and commitment, and training and ongoing support):

Team capacity and commitment

Some therapists thought the effectiveness of DBT was solely due to its techniques and theory [29], but this view was not universal. Several optimal therapist attributes were detailed including a stance of equality, an ability to synthesise validation and challenge, a good understanding of DBT skills, as well as, group management and teaching abilities [25]. Therapist confidence also seemed important [27] and this could be enhanced through DBT implementation [24, 28] and training [35]. Administrators selected staff based on their seniority and motivation and recruited to ensure team diversity [36]. Therapists’ academic qualifications seem less important [26], but they do need to be skilled clinicians [31].

Insufficient staffing can jeopardise sustainability ([23, 30, 36]) and staff turnover is a further barrier [23, 31]. For instance, in one prospective implementation study 55% of therapists remained working at their original organisation at two year follow-up [35]. A possible reason for retention problems is that new DBT therapists initially reported increased stress levels and there was a tendency (although this was not statistically significant) for staff burnout to occur over time [28]. A small association suggested that larger teams implement more DBT components [26]. Smaller teams are likely to operate within larger services, with staff having additional roles. These nested programmes seem common as several therapists reported working in the DBT team part-time [28] and contrary to the Ditty et al. [26] findings, there were examples of successful nested teams (e.g. [56]) and teams dependent on part-time staffing (e.g. [52]).

Training and ongoing support

Clinicians from diverse disciplines can acquire a solid grounding in DBT through training [34]. Training facilitates implementation [31] and attending more training is associated with greater confidence and use of DBT [27]. For instance, training significantly increased the use of skills training, treatment targets, daily diary cards and dialectic strategies [35]. DBT knowledge also moderately correlated with all indices of training [34]. Unfortunately, limited feedback about training has been collected. Therapists reported that training enabled them to use DBT in their practice but they wanted more detailed instruction on how to perform specific interventions, such as conducting chain analysis of problem behaviour [23]. In one study, E-learning was most successful in increasing reported application of DBT [32]. In an earlier report, instructor-led training improved therapist self-efficacy and satisfaction but no method increased therapist skilfulness [33]. There was some evidence that training could improve clinical outcomes (e.g. [66]) but a lack of training was not always a barrier: graduate students with minimal training achieved 88% fidelity with DBT methods when facilitating skills groups [64].

On-going consultation is important [23, 36] and lacking access to a DBT consultation team can be an implementation barrier [27]. DBT consultation teams can help therapists achieve dialectical synthesis [25] and complement [25, 28] or compensate for lack of training [34]. Access to individual supervision is also important [26]: lack of supervision was the most frequently reported barrier to using DBT skills in one report [33] and in a UK survey [31], 34% of respondents said supervision facilitated the use of DBT. Limited feedback has been collected about supervision experiences: therapists reported that supervision increases both stress and coping [28].

DBT

Our initial framework identified four sub-elements related to the intervention (design quality and packaging, adaptability, complexity and costs). Our search and synthesis yielded sub-elements related to the design quality, packaging and costs:

Design quality, packaging and costs

DBT can be a complex therapy to implement: several DBT skills can be difficult to understand and apply [25] and trainers have reported that therapists have difficulty applying DBT’s behavioural components [34]. Aspects of DBT which seem important are its treatment contract emphasising shared responsibility [29] and its adaptability [24, 30]. For instance, despite some authors believing that DBT’s manual-based nature is important [29, 34], there were many examples of adaptations (e.g. [48]) with adjustments often altering how telephone skills coaching was provided (e.g. [67]). In the US, limited reimbursement is a barrier to implementing DBT [27, 36] and in the UK, 29% of survey respondents said that allocating sufficient finances to DBT delivery was an implementation facilitator [31].

Implementation process

CFIR separates the implementation process into sub-elements related to execution, engagement, planning, evaluation and reflection. It was not possible to dissect these individual components in the literature. However, there were two examples of clearly executed implementation plans [40, 46] and five examples of services forming teams to oversee the implementation process ([41, 49, 51, 57, 59]). A lack of an implementation plan can be an implementation barrier [31] but plans do not guarantee success. For instance, one study planned to introduce a number of resources (e.g., demonstration videos, an online forum and telephone consultation) to improve DBT adherence during implementation [23]. During the study there were no requests for consultation and in post-implementation interviews therapists did not refer to any of the available resources. This study highlights that providing resources alone is unlikely to promote implementation.

Discussion

This critical literature review synthesised the DBT implementation literature to refine a DBT implementation framework. The framework sufficiently captured the data and no new elements or sub-elements were required (see Additional file 4). However, some refinements were made to create a more parsimonious and relevant framework for DBT. For instance, coding indicated that some sub-elements were capturing similar data. For example, the sub-elements ‘individual characteristics’, ‘facilitator skills/ qualities’, and ‘recruitment and selection’ were re-conceived into a sub-element called team capacity and commitment. Additionally, some sub-elements arose infrequently in the literature (e.g., research and published guidance) and these were therefore omitted (the refined framework is illustrated in Fig. 3). However, we acknowledge that limited literature on an implementation barrier is not necessarily evidence that the barrier is not significant in DBT. For instance, cost may prevent both implementation and research meaning that the magnitude of barriers related to cost may not be sufficiently reflected in the framework, as the literature primarily reflects successfully funded work.

Fig. 3
figure 3

Revised DBT implementation framework. Key: PARIHS, 20041; Damschroder et al., 20092; Fixsen & Blasé, 20093, Swales 2010a, 2010b4

Full size image

The utility of transdisciplinary implementation frameworks, such as PARIHS [5, 6], is highlighted by these findings: elements primarily derived from existing frameworks effectively captured DBT implementation barriers and facilitators. The critical review process also proved to be useful in guiding the framework refinement and the synthesis of the literature. However, only 14 papers were retrieved that specifically focused on DBT implementation and this suggests that a DBT implementation framework may usefully underscore the most important considerations for DBT implementers.

The DBT implementation framework is a useful resource for DBT practitioners and service leaders who are planning (or overseeing current) DBT implementations. The synthesis indicates that implementers should consider the following recommendations:

  • When introducing DBT into practice clinicians and organisations should encourage the staff team to operate a benign approach to BPD and ensure there are good communication systems in place.

  • When establishing a DBT team, it seems important to recruit therapists with sufficient cognitive flexibility, whose personal qualities align with those espoused by DBT, such as, having a non-judgemental stance.

  • The DBT team will benefit from on-going supervision and consultation and therapists should receive adequate training.

  • Leadership support is beneficial and in situations where implementation is not pre-planned, a DBT champion can help.

  • It is beneficial for services to evaluate whether DBT needs adapting to suit their organisation.

Despite the apparent strengths of the DBT implementation framework, the limitations of the literature need to be taken into account. The framework’s generalisability cannot be ascertained as the reviewed literature only provided information about implementation in Western contexts and primarily reported on implementation in statutory outpatient mental health services. The most commonly retrieved papers were trial reports and implementing DBT in a research context may have significant differences from implementation in clinical services. Furthermore, the decision to include only published literature biased the review towards considering effective DBT implementations as most trials and programme descriptions reported positive results. In particular the literature reviewed, with the exception of a DBT implementation with Native Americans [43], cannot inform how DBT implementation is achieved with marginalised and particularly high risk populations, such as cultural minority groups. When more information about DBT implementations with such populations become available, the framework may require refinement.

The implementation papers reviewed also had methodological limitations. Most data was collected retrospectively and relied on self-reports of implementation success (e.g. [26]). Samples may not have been representative, for instance, the response rate in one study was approximately 14% [30]. Furthermore, survey instruments had not always been validated (e.g. [35]) and most quantitative data was correlational (e.g. [26]), so causation could not be inferred.

Limitations in the literature and framework provide opportunities for future research. It is acknowledged that interrater agreement when using the framework to code data could be improved. The current framework is sufficiently detailed for use by DBT practitioners and service leaders who are planning implementation, but in a research and academic context one next step will be to develop more precise definitions of some sub-elements. Although, the current literature cannot inform how implementation barriers and facilitators interact or how they are weighted in different contexts, a few tentative potential relationships warrant further exploration. For example, communication and contingency management might be particularly important in organisations providing team approaches, such as, inpatient services. Access to on-going support may be particularly important if staff have not received comprehensive DBT training. The complexity of DBT may only form a barrier if clients and staff have not been appropriately selected. To refine and further develop the DBT implementation framework in the academic context, another next step will be to undertake further research to explore these tentative ideas about how the framework elements interact and are weighted. To explore these relationships further, prospectively collected data will be needed as is planned in the EMPOWER research programme (NIHR Programme Grant: RP-PG-1212-20,011).

Conclusions

This review has explored the DBT implementation literature and developed a bespoke framework to inform future implementations. The literature synthesis has highlighted some important implementation considerations but prospective DBT implementation studies are now needed to explore the relative weighting of and relationships amongst these barriers and facilitators.