Study design and methods

We conducted a qualitative study following on a quantitative survey that explored the extent to which US hospitals implemented practices to prevent hospital-acquired infections.3 4 11 12 Survey responses were used to select a stratified purposive sample of 14 hospitals for in-depth semistructured telephone interviews (see table 1). These hospitals were selected for their potential to further our understanding of organisational barriers and facilitators in implementing infection prevention practices.4 13 We first stratified hospitals based on size (⩽250 vs >250 beds) and Veteran Affairs (VA) versus non-VA hospitals. We then chose hospitals to maximise variation13 with respect to using CLABSI, VAP, and urinary tract infection prevention practices; whether or not a hospital participated in a collaborative and selected academic and non-academic hospitals. As is ideal in qualitative research, we used findings as the study progressed to refine our sampling strategy, and moved to a case-based strategy that emphasised particular combinations of characteristics that would best illuminate mechanisms that led to high or low use of infection prevention practices, including new criteria such as barriers to adoption and degree of institutional support for change efforts.14 Subsequently, 6 of the 14 hospitals (see table 1) were purposively selected for site visits to more deeply explore preliminary themes. For example, one site engaged in a quality improvement collaborative but did not seem to have effective champions, whereas another site that also participated in a collaborative did have readily identifiable champions. We also included a large public hospital with relatively strong ties to an academic medical centre and a small hospital that was only loosely affiliated with an academic medical centre because we were interested in how the medical education process and turnover of medical residents and fellows may affect implementation. The institutional review board from the VA Ann Arbor Healthcare System approved the protocol for the phone and on-site interviews and each study site’s institutional review board approved study protocols for the on-site interviews.

Interview and site visit protocols

We conducted two to four phone interviews (29–92 min) at each of the 14 sites (38 interviews in total) between July 2005 and May 2006 (see table 1). At least two study team members conducted each interview. The first interview at each site was with an infection control professional, who, along with the other interviewees, recommended other individuals who had a key role in implementing at least one of the prevention practices under study. In some cases, we also interviewed staff not mentioned by our participants. We asked interviewees what practices their organisation used to prevent CLABSI, VAP, and urinary catheter-associated infections and to explain how these practices were implemented at their facility or why they were not using them.11 14 Interviews were recorded and transcribed verbatim.

We conducted 48 additional interviews during visits (four person-days at each site) to six facilities between October 2006 and October 2007 (see table 1). The site visits provided an opportunity to gather more in-depth information, fill gaps in our understanding, confirm and explore issues identified by the phone interviews, and obtain perspectives from a broader sampling of individuals, including senior executives, midlevel managers and frontline clinicians. Interviews were tailored to each hospital and interviewee. We analysed more than 950 pages of interview transcripts.

Data coding and analysis

We conducted data analysis using rigorous qualitative procedures.15 16 Analysis was ongoing throughout the study.13 Summaries were produced after each interview and members of the study team met frequently to discuss emerging themes. A preliminary codebook was developed based on the study conceptual framework: Roger’s diffusion of innovation11; along with new codes developed through review of the data. Because of the complex and dense nature of our data, we used a consensus approach to coding.17 18 Transcripts and codes were entered into NVivo7 software. Code summaries from the phone interviews, along with transcripts of site visit interviews, were further analysed using a group consensus process. Four broad themes arose inductively from the data that were further explored: active resistors and organisational constipators as barriers to change,19 clinical and administrative leaders’ roles in creating a culture conducive to improving infection prevention practices, which quality improvement approaches worked best in which organisations, and the influential role of local champions. Initial findings related to each of these four themes were debated and challenged for each case, while comparing each case to previously analysed cases.20 Case summaries were refined through team consensus.21 All four themes played pivotal roles in failed or successful implementations. This article reports results related to the role of local champions.

Champions and degree of implementation

We identified champions by asking interviewees to tell us about people who played a major role in implementing a practice and whether there were individuals in “senior management who helped get things done.” Champions were identified through consensus of the study team and were affirmed by two or more of the interviewees. We characterised the degree of practice implementation at each site through team analysis of information provided by interviewees. We considered spread of the practice (eg, whether it was implemented in all units or only the medical ICU), the degree of commitment by users22 and their likelihood of sustaining the practice.

Results

Early in the analysis, it was clear that characteristics of infection prevention practices influenced the number of champions involved in implementation. Based on these findings, we characterised the practices described in table 2 as a technological change, behavioural change, or hybrid change. Specifically, we defined technological change as those involving purchasing products or services or automation of production processes—for example, replacing traditional catheters for antimicrobial catheters. Behavioural change refers to practices involving new organisational structures or processes—for example, elevating the head of the bed for ICU patients. Hybrid change involves new technology plus significant behavioural change—for example, using maximum sterile barrier precautions requires purchasing larger drapes and significant changes in central venous catheter insertion practices.

Table 3 shows the varying degree to which technology-focused and behavioural change infection control practices were implemented in the study hospitals. Hospitals tended to broadly implement either technological practices or behavioural change practices, but not both. Hospital A broadly implemented technological practices but had only limited success implementing behavioural change practices. Three hospitals—B, C and E—tended to implement behavioural change practices but not technological practices. Hospitals D and F broadly implemented only one of the two technology-based practices (subglottic secretion drainage tubes at hospital D and antimicrobial central venous catheters at hospital F) and had limited success with the behavioural change practices.

Technology-focused change

Generally, a lone, intrinsically motivated champion led the change effort in hospitals where a technological change (subglottic secretion tubes or antimicrobial central venous catheters) was broadly implemented (see table 3). Respiratory therapists championed subglottic secretion tubes at two hospitals (A and D). At hospital A, a respiratory therapist made a business case and took an incremental approach to build support:

I…showed the total cost of savings for the facility…. The tube is more expensive…but…the hospital still saves huge amounts of money…if you’re going to talk about preventing a pneumonia [sic]. So I just suggested that if we start in the [operating theatre] with high-risk patients and have them in the units, then we wouldn’t have to replace all the tubes.—respiratory therapist (hospital A)

The respiratory therapist went on to say that the transition to the new tubes was easy because the insertion process did not change:

It wouldn’t require the physicians to have any new working knowledge of the tubes because the tubes basically look the same…physicians would intubate the same…I really believe it is a very easy transition that could have fantastic results…

However, the ability of a lone champion to implement even a simple technology-focused change can be tenuous. It was a long and bumpy road for this respiratory therapist because it was 2 years before changes in staffing finally brought receptive physicians who supported the change into key positions.

We saw two general sources of resistance working against the efforts of lone champions of technological change: (1) if the technology is sufficiently different from the current technology, a foothold for resistance could arise to block the change by triggering doubts about safety, efficacy or having to change procedures; (2) the new technologies, in all cases, were more costly, which often led to arguments to try other “less expensive” approaches first. For example, individuals at most, but not all, of the hospitals we interviewed regarded subglottic secretion tubes as being different enough to raise resistance against using them (eg, “it’s a stiffer tube”). In contrast, most, but not all, interviewees believed antimicrobial central venous catheters were a straightforward “swappable” technology.

Even having a physician champion does not guarantee success. Physician champions encountered the same types of resistance against using new technologies that non-physician champions faced. At hospital B, a physician championed the subglottic secretion tube but respiratory therapists would not use it. At two other hospitals (C and E), physicians tried but failed to win approval for antimicrobial central venous catheters in the face of a counterargument to do “less expensive” approaches first.

Behavioural change

For behavioural change, it took more effort for champions to gain consensus—a process that one interviewee described as “corralling cats”—in their efforts to get buy-in and commitment from staff across professions and units to broadly implement the change. As reflected in table 3, broad implementation of hybrid or behavioural changes generally required multiple champions: coalitions of managers and frontline staff who worked across professions and units. Two of the six hospitals (C and E) broadly implemented sterile central venous catheter insertion procedures and head-of-bed elevation, which require significant behavioural change in multiple units within the hospital. Another hospital (B) broadly implemented the sterile central venous catheter procedures but head-of-bed implementation was limited. Three hospitals had only limited implementation of the behaviour-change practices (A, D and F).

A comparison of hospitals A and C provides a stark contrast in the degree to which champions succeeded in broadly implementing hybrid practices. Hospital C focused on implementing behavioural change practices before technological changes. Intrinsically motivated champions (physicians and non-physicians) convinced colleagues across professions to work together to implement sterile central venous catheter insertion and head-of-bed elevation practices. They accomplished this in a setting characterised by highly effective working relationships among key stakeholders; these relationships were clearly described and confirmed by all interviewees. Table 4 provides sample quotations that elucidate the strong connections forged through healthy working relationships at that hospital.

Champions at hospital C developed highly effective strategies, in collaboration with other staff that crossed organisational boundaries and professions to implement sterile central venous catheter procedures and head-of-bed elevation. For example, the infection control professional and a physician leader energetically championed the sterile central venous catheter insertion practice, garnering support from top and mid-level hospital leadership. They started with a “gap” analysis to compare their own practices against Centres for Disease Control and Prevention recommendations and submitted a “crude proposal” that was funded by their corporate office. The hospital system subsequently made them an “alpha site,” advertising it as a “big initiative,” which also got their hospital’s president on board. The hospital’s president then opened the door for the infection control professional to present the initiative to the head of the medical executive committee and at several management meetings. A newly hired physician leader replaced the initial physician champion and actively championed the new practice with the infection control professional, helping overcome resistance, especially from surgical residents, by being visible during their orientation. The infection control professional described this multifaceted support:

there were certain physicians that didn’t think they needed to follow the practices but having…the [physician leader’s]…name on our checklist…to see that things were being done properly…if…[a nurse spoke] out to a physician…Lord knows what kind of retaliation there could be related to that…the [physician leader] would talk to them. We have a very, very strong personality in our medical director so that didn’t become a problem…—infection control professional (hospital C)

The attitude of anaesthesiologists and surgeons towards change at hospital C was unique. At the other hospitals in our sample, these specialties were singled out as being particularly resistant and unapproachable regarding new behavioural change practices. For example,

[Anaesthesiologists]…tend to, in my experience, use medications, follow practices based on their…small world. Because they have to control everything in a patient, they interact less with other people and so hear less opinions….For hours, you are not collaborative [in the operating room]. You are running a show. Surgery’s not conducive to necessarily teaching, you are literally focusing on that stitch, you’re not asking for opinions. Although an OR’s big, it really is a solitary job that everybody’s doing, compared to Medicine. You know, we constantly collaborate, collaborate [sic] is not considered a weakness.—physician leader (hospital D)

At hospital C, however, the infection control professional told the story of how the chief anaesthesiologist agreed to champion use of sterile central venous catheter insertion practices in the operating room:

“how do you approach this?” The management Christmas party over cocktails where you get the head of the Anaesthesiology and you say, “Do you know about the central line bundle thing we’re doing in the ICU…and did you know that these infections that we had over the last three months came from the OR?” And then, that’s how we got them…I saw him there and I just thought, perfect opportunity…my philosophy has always been…what if it’s your mother, your father…we always want the best care for those that we love and try to bring that point home to everyone.—infection control professional (hospital C)

In contrast to hospital C, poor interpersonal connections characterised the setting at hospital A, which made it challenging for champions to implement behavioural changes. Instead, technological changes tended to be implemented here. Table 4 shows illustrative quotations that describe the poor quality of relationships and impediments that arose among professions and across hierarchical levels. Champions faced strong barriers at this hospital. Despite the efforts of many individuals to implement sterile central venous catheter insertion procedures, even with support from an external collaborative, the site continued to have among the (self-admitted) worst infection rates in the collaborative.

Furthermore, the champion at hospital A was appointed by leadership rather than stepping into the role through intrinsic motivation. This champion was clearly beleaguered, a feeling shared by other staff:

It just gets to be overwhelming after a while and they’re all a little bit different…nobody really got direction. This, “you will do this, you will have some time to do this”, it was just sort of dumped on…the person in the MICU…but there wasn’t enough hours in a day.—infection control professional (hospital A)

A key physician leader admitted that they were “not implementing these practices the way we should.” This leader asserted that it was up to the “frontline” staff to champion the change. The approach here was to take a “command and control” approach, saying, “don’t talk to me again, make it happen, I don’t want to hear any excuses, good-bye.” But no one took charge “with any vigour.” Others with whom we talked had similar stories of pushing responsibility for implementing onto others, with no one stepping up to make it happen.

One exception at hospital A was a relatively new ICU director who collected compliance data and provided feedback using graphs of progress, which then motivated staff to continue their efforts. This leader provides a clear example of how, even in this challenging organisation, champions can make a difference within their own domain. By the time of our site visit, this leader had started a new initiative in one ICU to sustain and build on earlier, external collaborative-driven efforts to implement sterile central venous catheter insertion procedures. Staff were encouraged to get a cart assembled with the supplies needed to comply with the new practice, and they designed procedures for supplying those carts. Nurses were empowered to monitor physicians and residents. The ICU Director found champions within the ICU who wanted the change and promoted healthy connections among staff:

we made sure that amongst everybody, even the nutritionist, the pharmacist, the problems dealing with all the big, big issues we get a strong champion…there’s two…senior nurses, very good, very strong in terms of enforcing things…pick them just because of their interest and their roles…There’s one or two other attendings that I tend to also get on board very quickly…—ICU director (hospital A)

Discussion

Regardless of practice and setting, champions who were intrinsically motivated to step up to the challenge of implementing a new practice were most effective. Enthusiastic commitment to a change is associated, when strong enough, with championing behaviour.23 In healthcare quality improvement initiatives, physicians are often appointed to fill a champion role for changing clinical practice.24 25 However, we described one example of how a top-down “appoint-a-champion” approach failed; a finding that is in line with previous research.26 27

Change that involved a “swappable” technology could be achieved with a lone champion in some circumstances. Change that required a significant degree of behavioural change, however, required coalitions of champions working with staff across professional and organisational boundaries. Investments in technology are not necessarily simple or inexpensive but may not require the same level of staff engagement as new practices that require extensive behavioural change.8 The interaction of multiple champions allowed each to apply their limited time and energy in a way that fit their setting. For example, we heard about physician leaders actively helping to build support for new practices among fellow physicians both in their own and across units, putting resources and procedures in place to encourage compliance. This was particularly important when faced with active resistance or organisational constipators who create insidious barriers to change19 and to minimise the effect of those potential barriers in the organisation.28 Together, with frontline staff, they built coalitions of support for the new practice across units, professions and hierarchical levels.

The capacity for building partnerships for broad implementation was tempered by organisational setting. Highly connected organisations (one in which healthy professional relationships are encouraged and established) created an environment that encouraged champions to rise up and build partnerships that were needed to implement a new practice. Implementation floundered at hospitals where professions (eg, nurses, physicians) or individuals at different hierarchical levels (eg, leadership vs frontline) did not exhibit functional relationships, creating a particularly challenging barrier for champions. In these organisations, champions felt alone and beleaguered by yet another initiative. The difference in the number and quality of connections among champions and other stakeholders was striking between sites with and without broadly implemented behavioural change practices.

The importance of interpersonal networks has long been recognised in the diffusion literature.29 Some researchers assert, based on complexity theory, that relationship quality may be more important than qualities of individual champions.30 We saw clear and specific examples of how relationships influence implementation of key infection prevention practices. Sometimes champions have to work “under the radar.” Who champions are and how they generate support can be a sensitive process in toxic settings.31 Thus, specific actions for engaging key stakeholders need to be tailored to the setting.26

Our findings should be interpreted in the context of several limitations. Our goal in this study was to understand more deeply the role of champions in the success or failure of practice implementation rather than to generalise findings across all hospitals. Because the quality and validity of qualitative analysis rest in part on understanding as much about each case as resources allow, we chose to study six hospitals in detail rather than do less in-depth data collection and analysis on a larger sample of hospitals. Nonetheless, our findings have application outside the study sample for those who recognise some of the same characteristics in their own setting and thus may provide insights into how to implement change in their setting.13 We have also generated hypotheses for further study.

Conclusion

In limited cases, simple technology changes can be implemented through the efforts of a lone influential or persistent champion. For practices that require significant behavioural change, however, a coalition of champions may be needed; although behavioural changes themselves may appear deceptively “cheap” compared with purchasing new technologies, the complexity of implementing changes in behaviour increases significantly because behavioural change usually requires building effective interprofessional coalitions. Even in poorly connected environments, however, individuals can champion change within their own sphere of influence.