Consistent with our theoretical model, results support the role of management support, resources, implementation policy and practices, the degree to which MOVE! fits with values and existing programs, and the implementation climate as antecedents to MOVE! implementation. The importance of the role of champions was mixed.
Administrative and clinical management at the regional and local levels actively supported efforts to implement MOVE! at the facilities with high implementation effectiveness. The importance of management support has broad backing in the literature . This finding is directly confirmed by a study of another widely disseminated complex program in VA aimed at reducing waiting times at primary care and specialty clinics in which a quantitative measure of management support was one of only two significant predictors of effective implementation . In Klein, Conn, and Sorra's implementation model, management support has a direct effect on implementation policy and practices. The more committed managers are to the implementation, the more likely they will provide the resources and support needed e.g., training, space, communications. Our data support the existence of this relationship. Managers at the high-implementation and transition facilities helped to establish supportive practices and infrastructure by increasing program visibility, dedicating staff time to MOVE!, making it clear to service chiefs that participation of their staff in MOVE! was expected, and provided active and moral support to frontline staff.
More active management support could be encouraged by explicitly defining roles and responsibilities for different levels of management to implement a complex program like MOVE!. In the absence of management support, local champions could help fill the void to some extent . We described one example of a physician champion at a facility with low implementation effectiveness who together with the above-and-beyond efforts by the MOVE! coordinator, succeeded in launching a program in a challenging context. It was striking that at three of the five facilities, little or no mention was made of their physician champions, even when asked directly. Thus, the importance of role of champion was mixed and may be compensated for by the positive presence of other constructs. It appeared that some of the physician champions were appointed into the role by facility leadership. It is important, instead, to identify champions who truly believe in the program, are open to change, are respected by their peers, have good communication and leadership skills, and willingly volunteer to take an active role by promoting the program broadly in their organization, serving as a local expert, using their influence to persuade peers to support and engage with the program - all this for the duration of time it takes to sustain skilled, enthusiastic use of the program [30, 31].
Four of the five study facilities provided initial assessments and a program of group visits but none of the facilities provided consistent phone-based self-management support, which is a foundational treatment component of MOVE!. Staff at our study facilities did not have sufficient time nor did they feel confident in their ability to help Veterans over the phone. As a result of this latter finding, NCP is piloting a national call center to provide self-management support for Veterans. In addition, a home-based "TeleMOVE" program was implemented in FY2010 that allows Veterans to receive daily motivational messages through a home-monitoring device, "checking-in" with weekly weights, and requesting a call from program staff when needed. These alternative programs overcome facility barriers related to space, logistical and transportation barriers to patients attending facility-based programs, and in the case of TeleMOVE, allows clinicians to monitor a large patient panel with an automated intervention.
The influential role of some implementation policies and practices, as defined by Klein and colleagues , was supported by our findings. The most frequently mentioned need was dedicated time. Staff at the two facilities with low implementation effectiveness did not have explicitly dedicated time allocated to MOVE!. At the transition facility and facilities with high implementation effectiveness, winning formal approval for dedicated time was a key enabler of their success. Once dedicated time was approved, the MOVE! coordinator, especially, was able to do the legwork necessary to negotiate for space and other resources needed to implement the program. Our qualitative findings also revealed the critical role communication played in making MOVE! visible within the organization . In addition to specific communications about MOVE! however, there is also support for expanding the definition to include the role communications play in coalescing teams and building a shared vision . MOVE! staff at high implementation facilities described a more cohesive team of committed staff, despite the fact that each team member reported to a separate supervisor in separate units. There is support for coalescing multi-disciplinary teams  contributing to successful implementation . These teams displayed an assertive, problem-solving approach to overcoming issues. This was in contrast to staff at the low-implementation facilities who did not meet face-to-face but rather communicated mostly through email.
Implementation climate, reflecting the importance of implementing MOVE!, was higher at facilities with a high level of implementation effectiveness. We described the challenge of implementation in a facility struggling to reduce large backlogs in primary care clinics which was a higher priority goal for the organization than implementing a new weight management program. VHA had implemented Advanced Clinic Access which is an established set of principles designed to give Veterans access to healthcare when they need it . Pressure to meet performance measures related to backlogs in primary care clinics was strong. We heard how, at one of the facilities, providers were working through lunch and on weekends to get their backlog of patients down. This highly visible initiative reduced the priority placed on MOVE! implementation. Other VHA performance measures related to reducing physiologic measures among patients (e.g., blood pressure) in FY2007 also seemed to work against MOVE! implementation efforts at one facility with low implementation effectiveness. Clinical leaders did not believe lifestyle change, as embodied by MOVE!, would impact these measures and thus were not willing to allocate the resources necessary to implement the program. A performance measure was implemented to promote screening for overweight/obesity and referrals to MOVE! at the start of FY 2009 . Over the course of two years (FY2009-2010), the number of patients screened and asked about their willingness to be referred to MOVE! increased from 79% to 96% system wide. This increase is likely to have increased priority for implementing MOVE! at local facilities . Establishing clinical priority for the program is also important. Providers at the two high-implementation facilities understood and articulated the role of MOVE! in helping their patients lose weight as a strategy to reduce the incidence or severity of obesity-related chronic conditions.
The qualitative findings supported our quantitative data, both of which validated our theoretical framework by highlighting significant differences in the contextual factors between the high and low implementation facilities. Management support, implementation policies and practices (communications and sufficient time), the fit of MOVE! with values and existing programs, and implementation climate were all qualitatively different and quantitative measures of these constructs were rated more positively at the high compared to the low implementation facilities. Furthermore, scores for the transition facility fell between low- and high-implementation facilities, reflecting the initial difficulties they had in getting the program implemented tempered by their more recent success in getting a viable program off the ground. The only construct that did not appear to distinguish between low and high implementation facilities was the role of physician champions, suggesting that champions alone are insufficient for overcoming other important barriers.
Thus, the quantitative measures appear to be reasonable indicators of the strength of influence of organizational factors in getting MOVE! implemented. The obvious benefit of the quantitative data is their ease and efficiency of administration, in contrast to the data obtained from the open-ended interview questions, which are particularly time consuming to code and analyze. Nevertheless, the qualitative data provide insight into exactly how each construct served as a barrier or facilitator to implementation effectiveness, thus providing the data necessary for making recommendations for future implementation efforts.
Several study limitations merit consideration and provide context for interpreting our findings. First, this was a cross-sectional study of a small number of purposively selected facilities and thus generalizability is limited. However, our goal in this study was to more deeply understand contextual factors facilitating or hindering initial implementation of MOVE! and how those factors differ between high and low-implementation facilities. A purposive sampling design was selected to maximize variation to identify important differences and potential common patterns across a diverse sample of medical centers. Second, the investigators, coders, and analysts together, assessed implementation of MOVE! treatment components and were thus aware of the status of implementation at each facility. This knowledge has potential to bias qualitative findings. However, the diversity of the team, its confirmation by quantitative findings, and our careful analytic approach helped to minimize this potential bias. Third, it is important to note that our findings were based on data collected in the second year after MOVE! was disseminated (18-22 months later) and do not reflect MOVE! implementation today. Lessons learned from this experience, however, can benefit other large dissemination efforts. Fourth, though we used a theoretical model to guide our data collection and analyses and our findings are promising, our sample of five facilities was not sufficient to validate the model for this setting. We had insufficient data to confirm the mediating roles of individual constructs in the model. Instead, we simply examined the presence or absence of each of the constructs, rather than their relationships with each other as depicted in Figure 1. Further research is needed across multiple studies to examine these relationships.