Our findings are organized by primary and secondary drivers. The four primary drivers that emerged during our review include Planning and Infrastructure; Individual, Group, Organizational, and System Factors; The Process of Change; and Performance Measures, and Evaluation (Figure 1). The findings for each driver are neither exhaustive nor mutually exclusive, and may need to be modified or reconsidered in particular settings, but they do point to general themes, ideas, and concerns that may have a general applicability to others involved in large-scale spread initiatives.
Driver No. 2: Individual, Group, Organizational, and System Factors
All large-scale improvement initiatives involve humans in complex social settings. Therefore, understanding the cognitive dimension of spread—how people and groups think about, interact with, and are affected by the innovation—is critical to developing, executing, and evaluating large-scale initiatives (Davidoff, 2009; Pawson & Tilley, 1997). Ferlie and Shortell argue for a four-level approach to change: individual, group, organization, and system (Ferlie & Shortell, 2001).
Individual and group factors
In considering how individuals engage with innovation, the theme of personal values emerges strongly in the healthcare spread and social movement literature (Bate, Bevan, & Robert, 2003; Bate, Robert, & Bevan, 2004; Della Penna et al., 2009; Dobbins et al., 2002; Ganz, 2008). Della Penna and colleagues, in their study of inpatient palliative care, write, “…the available literature on disseminating better practices in healthcare did not capture the sometimes intense sense of personal engagement that permeated all levels of the organization…broad personal engagement seemed to be a force stronger than individuals’ rational beliefs …” (Della Penna et al., 2009, p. 7). They identify three elements of engagement: a highly credible evidence base; a genuine belief that this new model provides better patient care; and an appetite for the innovation (“pull system”), rather than agents “pushing” the innovation to adopters (Della Penna et al., 2009). ExpandNet adds three factors: adopters perceive a need for and are motivated to implement the innovation; the adopting organization has the appropriate implementation capacity, decision-making authority, and leadership; and the timing is right (ExpandNet, 2009b).
A number of other individual factors affect adoption; a workload increase of 10% or more caused by the innovation may deter adoption (Bevan, 2010), whereas adoption is more likely if the adopter perceives that the innovation relieves workplace pressures (Bradley et al., 2004). Greenhalgh and colleagues challenge Rogers's (2003) five widely cited adopter categories (early adopter to laggard); they relate variation in individual adoption to general psychological antecedents (people's willingness to try new things), context-specific psychological antecedents (people's ability to use the innovation), the meaning of the innovation to the individual, the adoption decision, and how individual concerns during adoption are dealt with (Greenhalgh et al., 2004). Their study notes that the psychology of innovation and diffusion is not well researched within healthcare.
The literature notes the positive influence of people who model new behavior (change agents or clinical champions) and those who influence thinking about the innovation (opinion leaders) (Greenhalgh et al., 2004; Rogers, 2003), although their roles differ (Green & Plsek, 2002). For example, the managerial champion's role is to establish systems for implementation, whereas the clinical champion's role is to communicate technical knowledge about the innovation to colleagues (Nolan et al., 2005). Some clinical groups are more likely to adopt changes advocated by people from the same professional groups (Gardner et al., 2010). In all cases, providing champions and change agents with time to promote the innovation can maximize its benefit (Bevan, 2010). Clearly, managerial/clinical collaboration creates a synergy between subject matter experts, process owners, and stakeholders (Deming, 2000).
Strong leadership is consistently referred to in the literature as a key factor in scale-up and spread. Although charismatic leadership is suggested as an important factor in relation to large-scale social movements (Ganz, 2008; McCannon et al., 2008; Pastor & Ortiz, 2009), our analysis did not identify charisma alone as a sufficient driver of social change (Ganz, 2008). Indeed, the word “leader” is widely used by researchers without a clear definition.
Rooney and Leitch illustrate the key role that leaders can play in large-scale change, describing three factors that can move an entire country (Scotland) along in its quality journey: positive attitude; senior policy and delivery support; and a standard national improvement model (Rooney & Leitch, 2010). The work in Jonkoping County, Sweden, also highlights the importance of leadership (Ovretveit & Staines, 2007), where the persistence, commitment, continuity, style, and collaboration between the Chief Executive, Chief of Learning and Innovation, and Head of the Department of Medicine were considered pivotal in regional transformation. Several studies considered specific leadership actions to support change efforts—engaging staff, articulating the vision to the workforce (Della Penna et al., 2009; Gardner et al., 2010), identifying the target population, making the work a priority, committing time and resources to achieve objectives, and aligning organizational goals (ExpandNet, 2009b; Nolan et al., 2005).
In order to build a cadre of leaders equipped to effect positive change, the necessary leadership skills need to be identified and developed (Bate et al., 2003; Bate et al., 2004; Bevan, 2010; Bevan, Robert, Bate, Maher, & Wells, 2007; Ganz, 2008; Pastor & Ortiz, 2009). Bibby and colleagues suggest energizing leaders attract more effort and attract other high performing individuals, which leads to higher adoption rates (Bibby, Bate, Robert, & Bevan, 2007). Green and Plsek are the only authors in the scan to explicitly describe leadership competencies, which they tested by coaching “diffusion executives” (Green & Plsek, 2002) who achieved measurable improvement and faster spread than others.
Capability and capacity development
In addition to leadership development, institutional capacity and capability building are widely described in the literature as crucial to effective and sustainable scale-up (ExpandNet, 2009b). The highest performing healthcare organizations in the United States and United Kingdom have invested systematically in building improvement capability (Bevan, 2010), yet Bate and colleagues noted that fewer than 10–15% of the National Health Service (NHS) staff in England are actively involved in formal improvement activities at any one time. They estimated that to transform healthcare, 80–100% of NHS staff need to be involved in active improvement efforts (Bate et al., 2003). Research is needed to more thoroughly understand whether improvement capability exists and is underutilized (Bibby et al., 2007) or whether the capability gap is indeed this large.
Matrix identifies the following factors needed to support organizational/system capability building: provide access to appropriate skills training; create and embed specific roles with a remit for advancing the modernization agenda; and recognize the key role of middle managers in executing the vision and ensuring that frontline views are heard (Matrix, 2003). Bevan identifies seven skills to enable the workforce to improve healthcare quality and productivity at scale: process and systems thinking; personal and organizational development; involving patients, staff, and the public; initiating, sustaining, and spreading change; delivering on cost and quality; problem-solving/internal consultancy skills; and innovation for improvement (Bevan, 2010)—but does not suggest which skills are required at which level of the organization. Changes in formal undergraduate medical education to incorporate quality improvement training indicate a positive shift in equipping the next generation for large-scale improvement (Gould et al., 2002; Tsai, Bohnen, & Hafiz, 2010).
Learning and social networks
The literature widely supports the value of continuous learning networks (Matrix, 2003; O'Connor et al., 1996) to maximize workforce improvement capability ( Bate et al., 2003, 2004; Bevan, 2010; Della Penna et al., 2009; Dobbins et al., 2002; ExpandNet, 2009a, 2009b; Ganz, 2008; Green & Plsek, 2002; Greenhalgh et al., 2004; McCannon et al., 2008; Nolan et al., 2005; Pastor & Ortiz, 2009; Rooney & Leitch, 2010) and social networks to generate motivation and increase energy for improvement (Greenhalgh et al., 2004; McCannon & Perla, 2009). Structured learning environments, such as virtual or face-to-face training sessions are advocated on the premise that they bolster energy for improvement, develop a cadre of professionals skilled in implementation and spread, and shorten the timeline to full-scale implementation; however, it is difficult to assess the specific contribution of learning networks to achieving large-scale improvement (McCannon & Perla, 2009). Robust social networks aid knowledge exchange, create shared learning, and develop a sense of community among participants (Green & Plsek, 2002). Of course, not all social networks experience the same degree of success; networks work best in collaborative situations, where people feel comfortable sharing challenges as well as successes (Della Penna et al., 2009). The success of the network depends on a number of factors, including similarity of members; effectiveness of opinion leaders and champions in modeling adoption; recognition of achievements; working across organizational boundaries; and the presence of dissemination programs (Greenhalgh et al., 2004).
Organizational and system capability
The ability to plan and deliver widespread innovation requires scale-up and spread capability. Relationship building, leadership investment, and training are pivotal (Bate et al., 2003, 2004; Ganz, 2008; Mandel, 2010; Pastor & Ortiz, 2009). However, services or organizations with the greatest need for change often have the lowest existing capacity for introducing interventions (Gardner et al., 2010), whereas “cosmopolitan” (outward-looking) and “connected” (by proximity or shared networks) organizations (Greenhalgh et al., 2004) more readily adopt innovations (Dobbins et al., 2002; Greenhalgh et al., 2004).
In considering what healthcare could learn about spread from social movements, Bate and colleagues suggest a national structure to manage change through local “chapters,” whose leaders use tactics suited to local circumstance (Bate et al., 2004). Pastor and Ortiz found that large “anchor organizations” could lead smaller organizations in social mobilization (Pastor & Ortiz, 2009), as demonstrated by a national patient safety campaign that developed “node” healthcare organizations responsible for supporting and disseminating good practice through learning networks (McCannon et al., 2006).
Organizational and system culture
Sensitivity to local cultural norms and the ability to identify potential cultural strengths and weaknesses are critical components in planning and achieving large-scale change (ExpandNet, 2009a, 2009b). Positive cultural characteristics include enablement of cross-functional teamwork, support for pooled knowledge, creation of an urgent need to innovate, and sustained focus on the change (Gardner et al., 2010). Cultural behaviors that contribute to a culture of adoption and spread include internal and external surveying and assessment of healthcare innovations (Bevan et al., 2007), translating research into practice, and coordinating spread across all healthcare disciplines (Bradley et al., 2004; Green & Plsek, 2002).
Driver No. 3: The Process of Change
In addition to careful planning, leaders and managers of large-scale change need to select the process of change carefully. The literature identifies at least three dimensions: the extent to which the effort is actively pushed to participants, the underlying change theory that drives the work, and the mechanism used to spread the intervention. On the first dimension, spread can occur in various ways, from “let it happen” (natural diffusion) to “help it happen” to “make it happen” (active dissemination) (Greenhalgh et al., 2004). Nearly all of the scanned examples ranged between “help it happen” and “make it happen.”
On the second dimension, the underlying change theory, most change efforts used a clear model to drive the work; most common were the Model for Improvement (Langley et al., 2009) and systems thinking (Bevan, 2010; ExpandNet, 2009a, 2009b; Greenhalgh et al., 2004; Leape & Berwick, 2005; McCannon et al., 2007, 2008). Some programs encountered difficulties with the chosen theory, such as difficulty in rapidly testing change ideas using the “PDSA” format. It is unclear whether these challenges were related to the theory or to other structural factors, for example, lack of project team motivation (Vos, Duckers, Wagner, & van Merode, 2010).
None of the reviewed articles compared different theoretical approaches. Three studies identified the value of social movement thinking, including the use of powerful narratives, in healthcare (Bate et al., 2003, 2004; Ovretveit & Staines, 2007), and IHI's 100,000 Lives and 5 Million Lives Campaigns applied some aspects of organizing methodology (McCannon et al., 2006, 2007). It would be helpful to evaluate the contribution of the underlying change theory to large-scale change efforts and ways in which the various theories are complementary and contradictory (Bate et al., 2004).
The third dimension of the change process is the spread mechanism. Initiatives used a variety of delivery mechanisms, ranging from a small regional initiative that used data-sharing, learning, and site visits (O'Connor et al., 1996), to large collaboratives (Green & Plsek, 2002; Jha et al., 2003; Nolan et al., 2005; Schouten, Hulscher, van Everdingen, Huijsman, & Grol, 2008; Vos et al., 2010), to very large campaigns (McCannon et al., 2006, 2007; Slade, Tamber, & Vincent, 2003). Effectiveness of the change process varied widely. For example, collaborative methodology produced significant results in the VHA Upper Midwest Coaching and Leadership Initiative, where 26 teams in different organizations achieved measurable improvements in 17 different topic areas (Green & Plsek, 2002), but was unsuccessful in a program to redesign office practices and hospitals (Vos et al., 2010). Mittman argues that, given the widespread use of collaborative methodology, there is a paucity of balanced and systematic study on its effectiveness (Mittman, 2004). There are indirect references in the literature on the importance of matching the delivery mechanism to the problem. For example, a complex intervention that has not been fully tested might be more appropriate for collaborative methodology than a campaign structure. It is unclear whether the less successful change efforts were a failure of the mechanism or a failure to match the mechanism to the scope of the problem and the nature of the intervention. Evaluations of the delivery mechanism are increasingly complex, because many of the large-scale change efforts used hybrid or modified models (Gardner et al., 2010; Greenhalgh et al., 2004).
Developing a spread plan at the outset of a project can accelerate change by centering the team on the work to be done and identifying the desired end state. A number of studies, both evaluative and theoretical, outlined frameworks for large-scale spread and sustainability (Bibby et al., 2007; Bradley et al., 2004; ExpandNet, 2009a; Improvement, 2010; Matrix, 2003; Nolan et al., 2005; Pastor & Ortiz, 2009; Rooney & Leitch, 2010), but a clear synthesis or taxonomy of these frameworks does not exist presently.