Several authors have noted that the concept of facilitation has yet to be thoroughly defined and, thus, is challenging to integrate effectively into implementation research [3, 9, 16]. The QUERI reflective evaluation reported here, despite its limited scope, provides sufficiently suggestive findings from its multiple implementation experiences to highlight key aspects of the concept of facilitation – both in relation to the existing literature and the design of future implementation projects within and beyond the VHA.
Purpose, nature and role of facilitation
Interviewees from QUERI discussed the purpose of their facilitation efforts – actual and as desired in the future – in terms of interactive problem solving to meet specific implementation goals. This finding is consistent with Roger's definition of a change agent , as "an individual who influences clients' innovation decisions in a direction deemed desirable by a change agency." Therefore, it is different from facilitating a change that involves open-ended implementation goals and objectives.
The QUERI description of facilitation as a form of interactive problem solving also is consistent with a recently published concept analysis by Thompson, et al.  that identified the facilitator role as active, dynamic and task-oriented. Furthermore, interviewees' emphasis upon the criticality of the role's support element reinforces findings of both Harvey, et al.'s concept analysis that concluded "the facilitator role is about supporting people to change their practice [p.585,9]," as well as recent empirical work by Cheater, et al and Wallin, et al. .
In terms of the PARIHS definition and purpose of facilitation, there is an obvious overlap with this study's findings relative to the concept of a "helping and enabling" role [7, 8]. The heuristic definition that emerged from QUERI experiences further emphasizes what PARIHS would call the "task orientated approach" but provides more explicit operational detail about the "focused process of providing help and support to achieve a specific task [p.177, ]." Unlike the 2002 PARIHS description, QUERI facilitators at times did use "telling or persuading," given the goal of the external change agency.
In general, QUERI interviewees felt that facilitation could be a distinct entity, but questions still remain as to exactly how this problem solving and supportive function sets facilitators apart from other change agent roles, such as educational outreach/academic detail workers, opinion leaders, or champions. To some extent the findings from this evaluation support Harvey, et al.  in that QUERI interviewees described their facilitation role as broader than the educational outreach model; i.e., because as facilitators they engaged in more, and differently targeted interventions. For example, interviewees described their role in diagnosing and modifying contextual factors, and in being a boundary spanner/intermediary between organizations and other stakeholders. Arguably, educational outreach workers tend not to focus their efforts on context or on providing a mediating function.
Additionally, there may be some overlap between facilitation and project management. While participants in this study viewed their facilitation role as distinct from a project management role, in reality there may at times be blurring of role boundaries and tasks undertaken. According to the findings of this evaluation and those of Harvey, et al. , the distinction between a facilitation intervention and project management role seems to be one of intention and scope. A facilitation intervention, for example, is concerned with enabling the implementation of evidence into practice using a wide repertoire of skills and a flexible approach to working with individuals and teams in an enabling way. On the other hand, project management is not necessarily about enabling the process of evidence implementation and is potentially more restrictive in its scope, remit and enactment. More research is obviously needed to determine the specific differences between facilitation and project management roles.
In Thompson, et al.'s concept analysis , various roles of "intermediaries and influentials" in the transfer of knowledge were reviewed, revealing persistent confusion regarding both standard definitions of individual roles and cross-role similarities/differences. In terms of the roles reviewed – i.e., opinion leaders, facilitators, champions, linking agents and change agents – Thompson, et al. concluded that these specific "concepts may indeed be similar phenomena with different labels [p. 691, ]."
A facilitator in the QUERI study clearly can be defined as a change agent , as well as a linking agent between the internal change agent and the broader environment. However, as Thompson, et al. also found, "there are...many differences that suggest that these [reviewed] concepts are conceptually unique [p. 691, ]." For example, external QUERI facilitators did not conform to various descriptions of either an opinion leader (i.e., an internal individual who is informally well-connected and has a wide peer and social network) or a champion (i.e., an individual who emerges unsolicited and has visionary qualities).
A number of individual factors were highlighted by interviewees believed critical to the success of a facilitator's role. This included flexibility, relevant experience (as facilitators or as clinical process experts within the VHA), knowledge (e.g., regarding the evidence), and the ability to build relationships through good communication. The facilitators in this study were required to be flexible by adopting different styles (e.g., directive and non-directive) depending on the projects, specific sites, related progress, and individuals involved. These findings support others' work. For example, Cheater, et al.  found that flexibility and the ability to draw on a repertoire of skills supported the functioning of the facilitation role in an exploratory trial. Additionally, Wallin, et al.  describe the positive benefits of guideline implementation facilitators having knowledge about managing change, as well as having insight into the clinical topic, which they call being 'content aware.' Furthermore, Greenhalgh, et al.'s observations regarding external change agents  suggest that facilitators need to have credibility and be appropriately "trained and supported to develop... interpersonal relationships with potential users [p.26]." Thompson, et al.'s review generally echoes these observations .
In summary, and as Harvey, et al.  surmise, collective findings to date, including this evaluation, indicate that to be effective, facilitators need to be flexible and possess a range of skills to be used according to the needs of those with whom they are working and the related context.