From what we know to what we do: translating stroke rehabilitation research into practice


  • Conflict of interest: None declared.

Correspondence: Marion F. Walker, Division of Rehabilitation and Ageing, Queen's Medical Centre, University of Nottingham, Nottingham NG7 2UH, UK.



Despite the recent advances in stroke rehabilitation research, the translation of research evidence into practice remains a challenge. The purpose of this article is to communicate practical experience and describe research methodologies used to promote change and implementation of stroke rehabilitation research in three international settings. In England, the development of an evidence-based consensus document, combined with qualitative and quantitative methods, was used to promote practice change in community-based stroke services. The Canadian research program involved synthesis of evidence, creation of user friendly information, and development of multimodal knowledge transfer strategies to promote change at an individual clinician level. Australian researchers followed a multistep process, involving audit and feedback, identification of barriers, and tailored education to improve implementation of one clinical guideline recommendation. Reducing the evidence–practice gap requires the development of active management strategies. This article highlights the importance of close collaboration between stakeholders – both in terms of the transfer of evidence into clinical practice and for optimizing future Phase IV implementation research endeavours.


The evidence base for stroke rehabilitation has grown exponentially in the last 20 years. This step change may be attributed to an increased awareness that robust research findings are required to inform the successful delivery of rehabilitation practice. In addition, where there was once no research culture, there are now emerging opportunities for a career in stroke research for rehabilitation professions. What has been less impressive is the success with which research findings are actively managed by researchers and acted on within the clinical community.

Increasingly, it is recognized that implementation of research findings into practice needs a much more active strategy than simply disseminating findings in journals and at conferences. For example, in recent years, we have seen the development of Stroke Rehabilitation Guidelines, with individual countries compiling their own version of recommendations for best practice based on systematic reviews of the current evidence for specific assessment and interventions [1-4]. Although these guidelines have provided clinicians with a better understanding of the global evidence base underpinning stroke rehabilitation, dissemination of these documents is not enough to initiate and sustain a change in daily stroke rehabilitation practices.

The need to actively manage the movement of evidence into practice has also been recognized by the UK Medical Research Council framework for the development and implementation of complex interventions [5]. In this framework research moves from theory (preclinical) to modeling (Phase I), to the stage of exploratory trials (Phase II), then a definitive trial (Phase III), and finally, long-term implementation (Phase IV). This framework has helped rehabilitation researchers to better understand the continuum from generating evidence through to implementation. Unfortunately, there has been a paucity of Phase IV implementation research that is specific to stroke rehabilitation. There are few practical examples in the literature to help the stroke rehabilitation community learn how to move what they know to what they do.

Thus, the purpose of this article is to highlight international examples of Phase IV work being undertaken by stroke rehabilitation professionals. The authors form a collaborative group of researchers who wish to share their developing expertise with the wider stroke rehabilitation community. A range of research methodologies applied to address different evidence–practice gaps are presented. Some of the studies are ongoing, and results are not yet available. However, by highlighting the methods and interventions used, this information may be helpful for readers who are designing their own implementation studies. The terms ‘implementation’ and ‘knowledge translation’ are used interchangeably, because terminology varies across countries. To help readers compare approaches across countries, we have used the Knowledge to Action model created by Graham et al. [6]. This model shows key elements of the implementation or Knowledge to Action process (Fig. 1). The model includes a ‘Knowledge Creation’ funnel, representing the major types of knowledge or research that exist and can be synthesized to inform health care. The model also features an ‘Action Cycle’ representing activities that may be needed for knowledge application. The various steps presented in the model, such as adapting identified knowledge to the local context, assessing barriers to using knowledge, tailoring implementation interventions and evaluating outcomes, are featured in different ways in the three examples that follow.

Figure 1.

The Knowledge to Action model by Graham ID et al. [6].

Used with permission: Copyright © 2006.The Alliance for Continuing Medical Education, The Society for Academic Continuing Medical Education, and The Association for Hospital Medical Education.

Implementing evidence-based early supported discharge (ESD) services in England

In recent years, the rehabilitation of patients with stroke has become a national priority in England. Key policy and clinical guideline documents have recommended the implementation of stroke ESD services: teams that facilitate transfer of care of stroke patients from hospital to home and provide rehabilitation and support for patients at home [2, 7]. Recommendations to provide ESD are supported by robust research evidence that demonstrates that ESD is effective. Appropriately resourced ESD services provided for mild to moderate disability can reduce long-term dependency and admission to institutional care as well as reduce the length of hospital stay [8, 9]. In trials in which an economic analysis has been carried out, opportunity savings from hospital bed-days released tended to be greater than the cost of the ESD service [8].

What remains unclear is whether the health and cost benefits of ESD are still evident when services are implemented in practice. Despite recommendations by national clinical guidelines and robust research evidence, a recent Care Quality Commission report [10] indicated that many regions in the United Kingdom were still without ESD services. In addition, some ESD services that have been commissioned have failed to capitalize on conclusions drawn from research papers and bear little resemblance to the types of services advocated.

Our approach has been to focus on organizational issues associated with evidence-based practice. This program of work addressed four main questions.

What are the key research messages that require implementation?

Although national clinical guidelines in the United Kingdom recommend ESD for a certain group of stroke patients [2, 7], they lack information on how to implement services in practice. Having identified this lack of knowledge as a gap or problem to be addressed, our first aim was to establish the key messages from the research literature that we wished to implement. This step is analogous to the ‘Knowledge Creation’ phase of the Knowledge to Action process [6].

First, we were mindful of the fact that purchasers and providers of community stroke services probably do not have time to read long academic papers or systematic reviews. An accessible, easy-to-read document was required that summarized key research messages from the literature. A second issue was that conclusions in research papers usually do not directly translate into clinical practice. Authors of academic papers report on a basic research question: for example, whether a policy of early hospital discharge with support could be as effective and efficient as conventional care. What we felt was needed was an expert interpretation of results, ideally building on a systematic review of a number of randomized controlled trials and informed by clinical experience. To address this challenge, we used a modified Delphi approach and developed a consensus implementation document [11, 12]. Consensus statements were developed from the 2005 Cochrane systematic review [8] and specifically related to implementing ESD services in practice. These statements were reviewed and consensus reached by an expert panel of ESD trialists in a multistage iterative process. This process permitted the researchers to clarify the key elements of an ESD service and identify the issues that remained unresolved by the research literature. The published consensus document [13] has been disseminated internationally and has informed ESD service specifications throughout England.

What is the context in which the intervention is going to be implemented?

The second step in our program of research was to use qualitative techniques to explore the context within which ESD services were being set-up locally. Again, in keeping with Graham's Knowledge to Action model [6], we explored how the knowledge created in the ESD consensus document [13] could be used locally. Through semi-structured interviews, challenges and successes experienced by local ESD services were explored in order to increase our understanding of issues that service providers and purchasers were facing on the ground. We explored both individuals and their behaviors and wider political, financial and organizational factors [14]. Key themes identified included decision making involved in determining eligibility of stroke patients for ESD, establishing the intensity and duration of the intervention delivered from both a clinical and procurement perspective and training requirements of a stroke specialist, multidisciplinary team. This qualitative approach also helped us to determine if the models being used by the local ESD teams were informed by research evidence, and if not, why not. The interviews allowed exploration of topics about which consensus could not be reached and for which definitive answers had not been found using randomized controlled trial designs. This approach provided the ‘flesh on the bones’ and complemented the core elements of ESD services defined by the consensus.

How can implementation of the intervention be facilitated?

With both consensus and qualitative research findings, it was then possible to design activities to facilitate evidence-based implementation of ESD. Our approach incorporated strategies that have, in other settings, improved the provision of evidence-based care [15, 16]. In addition to organizing Knowledge Exchange [6] events, we have piloted tailored, team-based workshops designed to address key themes identified in the qualitative research phase. Our strategy was to cultivate a collaborative and symbiotic relationship with stakeholders [17] and the research program supported ‘diffusion fellows’; clinical champions employed one day per week to work with the research team. Having a position of clinical leadership, these individuals acted as a conduit for engagement and influence within partner clinical organizations [18].

It was not our intention with this program of work to formally evaluate the effectiveness of the implementation strategies used (‘Evaluate Outcomes’ step of the Knowledge to Action model). Instead, this pilot work was required to identify the best methods for implementing ESD services in practice.

Is the intervention still beneficial in practice?

The final strand of the research was to use quantitative methods to evaluate the outcomes of ESD services when implemented in practice. Our research question was: are the benefits of ESD, identified in randomized controlled trials, still evident following implementation? For ESD services, there may be reasons why patient outcomes might not match those achieved in the original trials. As outlined above, the model of ESD service in operation may not be consistent with what research findings advocate is most effective. Of more concern could be that patient variability, organizational challenges, and political tensions that come into play when services operate in the real world may mean benefits are no longer attainable. Our aim was also to identify optimal approaches for measuring the effectiveness of ESD services operating in practice.

Although randomized controlled trials are the gold standard in clinical trial design, it has been recognized that there is a need for additional quantitative approaches that can be conducted in less controlled clinical settings [19, 20]. The researchers therefore used a quasi-experimental approach, following two naturally occurring groups – an ESD cohort and a group of patients of similar stroke severity, who were not able to access ESD (usually due to lack of capacity in the service or a postcode lottery). Standardized outcome measures, as used in the original trials [8], were used to monitor patient functional ability over time. Data were also collected about the care pathway that each patient received. This outcome study is still ongoing and will permit measurement of the impact of ESD not only on patient recovery but also from an economic perspective by monitoring hospital length of stay and use of other community services following discharge from hospital. Data obtained to date have illustrated the complexity of the stroke care pathway operating in practice.

In summary, the aim of this program of research was to promote the implementation of evidence-based community ESD stroke services. The researchers were interested in both the successes and challenges of translating research findings into practice.

Stroke rehabilitation implementation of an individual-focused intervention within the Canadian context

In Canada, as elsewhere internationally, there is now a wealth of research on which to base decisions regarding the best practice management of an individual who experiences a stroke. In recent years, we have created a research agenda aimed at increasing best practices in stroke rehabilitation. This agenda has been based largely on the ‘Knowledge to Action’ model created by Graham et al. [6] for improving best practices at the local as well as national level. This model provides a sequence of phases for researchers and clinicians to follow in order to optimize knowledge transfer (KT) and to incite change in patient care [21]. If we refer to the ‘Identify Problem’ phase of the ‘Knowledge to Action’ model (see Fig. 1), it indicates the need to first understand the current gap between best practices and actual practices before undertaking KT initiatives.

To assess the extent of the problem specific to stroke rehabilitation, the research team conducted the Canadian National Survey of Stroke Rehabilitation Practices [22-25], which investigated the patient management practices of 1800 stroke rehabilitation clinicians working in stroke care. This study revealed gaps that can be summarized globally as – there was evidence of great variations in practice with a high prevalence of use of noneffective interventions and a less than adequate use of interventions with demonstrated effectiveness. For example, we found that less than 30% of Canadian occupational therapists employed best practice assessment and only 58% offered any intervention for unilateral spatial neglect (USN) – a serious sequelae of stroke that has specific best practice assessment and intervention guidelines [26].

Working toward a synthesis of the stroke rehabilitation evidence

In the years that followed, the research group conducted focus groups and interviews with researchers and clinicians regarding the possible reasons behind the problems/gaps that had been observed in the Canadian National Survey of Stroke Rehabilitation Practices. Clinicians suggested that they were having a difficult time incorporating the large amount of new evidence into clinical practice and suggested the need for a synthesis of the information on assessment tools and effectiveness of specific stroke treatment interventions (e.g., constraint induced movement therapy and functional electrical stimulation). From this feedback, we identified a need for the creation of user-friendly information that pertained directly to assessment and intervention and that could be applied in specific clinical settings. This identified need is in keeping with the argument put forth in the ‘Knowledge to Action’ process regarding the importance of ‘Knowledge Synthesis’ and the creation of ‘Knowledge Tools/Products’ (see central triangle in Fig. 1) to foster easier knowledge translation. An international group of stroke researchers came together and received funding to create two web-based tools that provide synthesis of stroke rehabilitation specific information aimed at clinicians working in the field (see;

We were aware that before initiating KT implementation studies directed at closing the gap between best practices and actual practices, we first needed to gain a better understanding of the various strategies that could be most effectively used with rehabilitation clinicians. Toward that end, we undertook a systematic review on the effectiveness of various KT strategies specific to rehabilitation professionals [27]. The evidence suggested that multimodal active educational methods to enhance best practice knowledge use by rehabilitation clinicians such as opinion leaders, interactive learning programs, training on evidence-based treatments and measures, role playing, and follow-up contacts were more effective than a single passive educational method such as an in-service or dissemination of guidelines.

Designing and testing KT intervention studies aimed at rehabilitation clinicians

As we continued our research agenda, we moved forward to the ‘Select, Tailor, Implement’ Phase of Graham et al.'s model (see Fig. 1). To begin, we designed a small study where we first held focus groups specific to the desired types of KT strategies that clinicians would find helpful [28]. Clinicians indicated numerous needs, including for example easy to access ‘pocket-cards’ with summary information they could use ‘on the go’. These suggestions, along with the evidence from the systematic review [27], were used to create a stroke specific KT intervention. Specifically, this intervention aimed at increasing clinicians' knowledge and self-efficacy in the management of poststroke unilateral spatial neglect. The intervention was then pilot tested. Specifically, the objective was to evaluate the feasibility and preliminary effectiveness of a multimodal KT intervention geared toward increasing evidence-based practice knowledge acquisition and self-efficacy for USN assessment and treatment. We selected the topic of USN given, as mentioned earlier, that we had strong evidence of a gap between best and actual practices [26]. Acute care occupational therapists treating patients with poststroke USN participated in two preintervention assessments, a day-long interactive multimodal KT intervention and a subsequent eight-week follow-up. This was followed by a postintervention assessment. Knowledge of evidence-based problem identification, assessment and treatment of USN, and self-efficacy to perform EBP activities were measured using standard scales.

The results indicated that all 20 participants completed the full protocol suggesting feasibility of an intensive KT intervention. A significant improvement in knowledge of best practices USN management and self-efficacy in carrying out evidence-based practice activities was found [29]. The findings suggest that using a multimodal KT intervention was feasible and could significantly improve knowledge of best practices specific to stroke as well as improve clinician self-efficacy for evidence-based practice.

Although the use of an intensive KT intervention such as this is desirable, most clinicians will be unable to participate in more than a few implementation programs of this intensity. Practically speaking, KT interventions need to be usable for a large number of clinicians, when the clinician has the time and the information needs to be updated easily when new evidence emerges. Web-based KT interventions fulfill these requirements. However, because they are usually not multimodal when used alone, they may not be effective in changing clinician knowledge. Our next objective was to address the effectiveness of a KT implementation using a web-based resource. Thus, in our next study, we examined the extent of knowledge acquired regarding stroke best practices by clinicians after using two web-based resources – StrokEngine-Intervention and StrokEngine-Assess. A random sample of 327 stroke rehabilitation clinicians in Canada accessed StrokEngine during a three-month intervention period. They responded to standardized questions about their clinical management of a typical patient with stroke (depicted in a patient vignette) during two baseline sessions and at one-month intervals during the intervention period. Data are currently being analyzed. Findings should help determine whether a web-based resource – used in conjunction with patient scenarios depicting actual patients with stroke– can improve best practice stroke rehabilitation knowledge.

Implementing an outdoor journey intervention in Australia using audit and feedback, barrier identification, and tailored education

Audit of medical records and feedback are often used to highlight an evidence–practice gap or problem (see ‘Identify Problem’ phase of the Knowledge-to-Action model, Fig. 1). The process of conducting an audit on its own is unlikely to change practice. However, when feedback is combined with audit, this pairing of interventions can be a powerful force for change [30, 31]. Feedback can create urgency leading to modest changes in practice. For that reason, several international health services and organizational bodies, such as the National Stroke Foundation in Australia, conduct biannual audits of stroke patient care [32].

Audit and feedback have been used to change stroke patient care in several Australian studies [33, 34]. In most studies, feedback about baseline audit findings is followed by a process of barrier identification (the ‘Assess Barriers to Knowledge Use’ phase of the Knowledge-to-Action model). Tailored education can then be provided to target known barriers. Thus, a complex package of intervention – audit and feedback, barriers identification, and education – is used across multiple teams (‘Select, Tailor, Implement Intervention’ phase of the ‘Knowledge-to-Action’ model).

One method of evaluating the effect of complex interventions on practice is with a repeated measures, before-and-after design, to explore feasibility, followed by a more robust cluster randomized controlled trial design (c-RCT). With c-RCTs, stroke units or teams are randomized to receive a tailored package of intervention, or written education such as clinical guidelines. This Australian example describes a feasibility study, which leads on to a c-RCT.

Australian guidelines recommend that stroke survivors should be offered a series of escorted visits and transport information from a rehabilitation therapist to help increase outdoor journeys [4]. Yet local medical record audits revealed that occupational therapists and physiotherapists documented very little about outdoor journeys and transport after stroke [34].We conducted baseline audits of 77 medical records across five posthospital stroke services as part of the feasibility study. These services included two outpatient programs, a day hospital, and two community home visiting services. Only 17% of the 77 people with stroke received six or more sessions of intervention targeting outdoor journeys, the ‘dose’ of intervention provided in the original trial by Logan et al. [35]. The medical record audit highlighted ‘underuse’ of an evidence-based intervention by stroke services, or an evidence–practice gap (the ‘Identify Problem’ phase). That feedback was communicated back to teams, at in-service sessions, and in a written report.

Barrier identification or assessment occurred after the baseline audit and after feedback had been provided to find out why services were not routinely providing evidence-based care (‘Assess Barriers to Knowledge Use’ phase of the Knowledge-to-Action model, see Fig. 1). To identify barriers, we conducted interviews and a focus group [36]. Questions were asked about knowledge (what did professionals know about the published research or how to deliver the intervention?), role expectations (did professionals expect or want to deliver the intervention?), and resources (such as whether vehicles, space, or money were available when needed). In the Australian feasibility study [34], we found perceived expectations of stroke patients, and their family was a barrier for therapists who wanted to provide escorted outdoor journeys. Therapists thought that stroke patients attending outpatient occupational therapy would expect to receive upper limb rehabilitation, not escorted outings. Therapists also thought that relatives would be reluctant for outings to occur soon after hospital discharge, in case a fall occurred. These expectations may or may not be true. They may not be an accurate reflection of patients' views and should be checked and challenged. Once barriers (and enablers) have been discussed, they can be actively targeted with tailored intervention. Education is typically a necessary part of intervention to promote evidence-based care.

Education was a third component of our intervention package, after audit feedback and barrier identification, to promote practice change across five health services (‘Select, Tailor, Implement Intervention’ Phase of the Knowledge-to-Action model). Therapists needed to understand what intervention had been provided in the published research and help to appraise the original research [37]. Communication was necessary between the researcher, service teams, and the author of the original trial in order to learn ‘what to do’ and ‘how to do it’. A half-day workshop was used to update knowledge, discuss and then target known barriers to practice change.

The final step in the implementation process was a repeat audit, 12 months after the first audit to measure practice changes (‘Monitor Knowledge Use’, see Fig. 1). In our study, a change of 15% was recorded – 15% more stroke patients received six or more therapy sessions aimed at improving outdoor journeys [34]. Some teams changed more than others. Qualitative interview data about barriers and enablers helped explain some of the between-team differences [36]. A c-RCT is now underway, testing the efficacy and cost-effectiveness of the implementation training package with 20 stroke care teams in Australia.

Summary and ‘where to from here’

The purpose of this article was to share our collective, practical experience in addressing evidence–practice gaps in stroke rehabilitation across three different international settings. We have outlined different approaches and methodologies designed to facilitate implementation and KT. Each example has targeted a different practice gap and followed steps described in the Knowledge-to-Action cycle. Furthermore, the examples selected relate to different service provision environments and highlight the need for careful selection of approaches to evidence translation. Our key message is the critical need for active management strategies to facilitate implementation.

The consensus statement approach used in the United Kingdom involved development of a consensus implementation document to address the gap in accurately translating research findings into protocols that could be practically communicated to and implemented by service organizations in the community. The document included expert interpretation of results that were easy to read [13]. This approach highlights the critical need to accurately operationalize the intervention with an appropriate level of detail and in a format that can be used by the service and health professionals. The implementation process was supported by semi-structured interviews and clinical champions and evaluated using a quasi-experimental cohort approach and standardized outcomes.

In comparison, the evidence–practice gap addressed in Canada focused on best practice management of neglect in individuals who experience a stroke and the need to create specialists that could provide the evidence-based intervention. The solution involved synthesis of evidence, creation of user-friendly information and protocols, use of multimodal active educational methods to enhance best practice knowledge use by clinicians, and creation of a intensive, multimodal KT intervention for the clinician [21]. A repeated measures before-and-after design evaluated the clinicians' knowledge and self-efficacy for evidence-based practice of neglect. This methodology permitted evaluation of expected change at the level of the clinician. More widespread education is being facilitated with web-based KT interventions.

In Australia, another repeated measures before-and-after study evaluated the outcomes of a KT package to increase escorted outdoor journeys by therapists. The KT intervention package included feedback following a medical record audit, barrier identification, and tailored education about ‘what to do’ and ‘how to do it’ to help promote implementation of guideline recommendations across five health services [34]. The next step is to conduct a c-RCT to test the efficacy and cost-effectiveness of the implementation training package. The cluster randomized trial with a concurrent economic evaluation is considered the gold standard evaluative design for the conduct of implementation research [33].

The need for Phase IV implementation research has been highlighted. Despite advances in the evidence base for stroke rehabilitation, translation to clinical practice has been slow [38], and the field of stroke implementation research is still evolving. How we can most effectively transfer research evidence into practice is still a challenge. Stakeholders, including clinicians, researchers, patients, services, and policymakers, need to work together to bring about change.

A number of opportunities for development and collaboration now exist. A coordinated approach to KT presents the opportunity to foster and develop a research culture in more clinical settings, perhaps through the involvement of champion therapists and joint clinical research positions. There are also opportunities for international collaboration to develop and pool resources and expertise and organize large-scale implementation cluster trials. In order to facilitate these goals, we have, in the first instance, formed a small group to promote best practice in stroke implementation research. It is proposed that an international consortium be formed that is inclusive and involves implementation researchers and stakeholders worldwide. It is proposed that the consortium provide access to implementation researchers willing to share their collective expertise in the field; set up links to resources developed to facilitate implementation research; and develop an international organizational structure to facilitate a coordinated approach to large-scale trials as well as links with related organizations and stakeholders. It is envisaged that the group would meet regularly via Skype and at international stroke meetings and seek funding to facilitate the group's activities and goals. We hope that the examples discussed in this manuscript will provide some initial direction in translating ‘what we know’ to ‘what we do’ for better outcomes in stroke rehabilitation.