Ethics

This study was granted ethical approval by National Research Ethics Service (NRES) Committee South Central—Oxford C (Rec reference number: 15/SC/0294) on 04.06.2015. All participants gave informed consent to participate in this research.

Aims and design

To assess the acceptability of PERFECT-ER and then to apply knowledge relevant to refine it, we used developmental action research [20–22] with an extended case study design [23–25]. This facilitated implementation data to be collected, for the iterative process of practice change in three acute orthopaedic wards in different NHS hospitals. We treated each ward as a case, using case study methods to assess the practices of each ward and then within- and between-case analysis across sites. The research team observed and interacted with each site as action researchers, collecting observational and documentary data over a 12-month period.

In each site, a Service Improvement Lead (SIL) used the Plan-Do-Study-Act (PDSA) model to lead implementation, for four PDSA cycles over 12 months (each cycle lasting three months) [26]. PDSA provides a model for structuring and encouraging change by recording and reflecting on implementation processes, actively learning about the intervention and its implementation, adjusting, and systematically sharing these lessons across multiple settings and time points [27].

Settings

This research took place in three English geographical regions selected to embody differences in locations on the rural-urban continuum. Table 1 summarizes information about participating wards.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. SIL and PL characteristics.

https://doi.org/10.1371/journal.pone.0279651.t001

Description of the intervention

PERFECT-ER is a flexible multicomponent service improvement intervention with an implementation model. The intervention is: the PERFECT-ER checklist (henceforth checklist) which synthesises best practice for hospital-based dementia care with current best practice for hip fracture care, change agents (SIL and PPL) and a model for change (PDSA). Checklists have previously been used in acute care to improve practice [28], however this is not currently in the public domain but described more fully elsewhere [29].

Change agents: Service Improvement Lead (SIL) & PERFECT-ER Process Leads (PPL)

In each site, trained SILs acted as change agents [30], using the PDSA processes whilst providing critical evaluations of the PERFECT-ER intervention to the study team for refinement. These SILs were nurses, seconded for three days a week from their usual roles. They were supported by a PERFECT-ER Process Lead (PPL) for an hour a week. These were senior hospital-based physicians or surgeons who provided mentoring and liaised with senior medical staff to facilitate implementation (see Table 1 for SIL and PPL demographics). All provided informed consent to participate.

Participants

NHS staff, including nurses, healthcare assistants (HCAs), physiotherapists and occupational therapists, provide daily care on orthopedic wards. Others, including orthogeriatricians, geriatricians, orthopedic surgeons, anaesthetists, pharmacists, and social workers, spend a proportion of their daily time on individual wards. We recruited any staff who regularly (at least weekly) delivered care on study wards, since they would be involved in delivering care relevant to the PERFECT-ER intervention and therefore involved in the implementation and action research process. Information sheets were given out by SILs and researchers visited to explain the study further and take consent. All participants provided informed consent (Site A = 80, Site B = 95, Site C = 95). Identifying personal and site features have been changed to maintain anonymity.

Processes and data collection

Each SIL performed a case note review of ten randomly chosen recently discharged patients with hip fracture and CI, to compare current practice with the XXXX checklist, then creating a score at the start of each PDSA cycle. These results were used to identify areas of strength and improvement. SILs had ward-based action-planning meetings, holding more or less meetings according to their implementation needs, to discuss the results then created and enacted action plans. Change was evaluated using the checklist to find evidence, or not, of service improvement. Action research data comprised: checklist scores from patient case note review, researcher field notes of action-planning meetings, field notes from teleconference meetings with the research team, and SIL generated documents including: their action plans and weekly reflective reports of activities undertaken.

Analytical approach

We used NPT to analyze and explain the work of agents taking a PDSA approach to implementing and embedding the intervention within different contexts over time. NPT describes four linked sets of mechanisms involved in the normalization of change [10, 11]. The first is ‘coherence’, the ‘sense-making’ work that enables actors to understand new ways of working. The second is ‘cognitive participation’, the work of building relationships so that actors can work together around a change initiative. The third is ‘collective action’ and concerns the interconnected activities of undertaking and sustaining new ways of working. Finally, ‘reflexive monitoring’ relates to the work of appraising the implementation process and informing change in future practices. NPT provides a robust framework for analyzing intervention implementation and is used here to inform comparisons between sites and over time. By considering temporality in the NPT framework, this paper offers new insights into how normalizing work changes over time to deliver sustainable implementation.

Analytical process

Data from across the sites and PDSA cycles were inductively analyzed by the research team (XX, XX, XX, XX, XX, XX). Following the identification of initial codes, this team collectively developed the coding strategy. Two researchers (XX and XX) used this strategy to code all data to the NPT constructs and PDSA cycles of change. This approach enabled researchers (XX and XX) to apply initial inductive codes to the NPT framework and then move back and forth iteratively between data and theory to identify implementation actions [31]. These were then examined by the whole team and key themes extracted for each NPT construct, site, and cycle, producing a representation of the change process in each site over time. In the next section we present findings for each site before integrating these, examining changes over time and across sites. All Names are pseudonyms to protect confidentiality

Findings in Site A

“…what I learnt from being here 10 years is that people don’t think of ways to improve things day-to-day but just come in and do their job and go home…” (Elizabeth, SIL Site A, Cycle 1, pre-action-planning meeting with researcher)

As an experienced nurse from an adjoining ward, Elizabeth (SIL, Site A) framed her involvement as an opportunity to get: “…back to [the] better orthopedic service they had before…” As illustrated in Table 2, whilst making sense of the intervention checklist, in cycle 1 Elizabeth took time to understand its contents, locating and assessing each item’s legitimacy within Trust policies and national guidelines. Trust policies dictated which items she should implement or not. Organizational support was provided by an ongoing working relationship with Site A process lead Marcus, who she described as “a very influential doctor” on the ward.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Normalisation process at SITE A.

https://doi.org/10.1371/journal.pone.0279651.t002

In cycle 1, Elizabeth’s early sense-making and reflexive monitoring focused on optimising the align of the checklist with the existing ways of working. In particular, she drew on her developed understanding of local practices to re-organise the checklist documentation and its storage prior to the first action planning meeting (see Table 2). At the first action-planning meeting, Elizabeth introduced the intervention, showing how selected changes fitted with Trust policies or national guidance, highlighting low scoring checklist items she wanted to improve. She showed colleagues how these changes would be implemented and what they would need to do to implement them. Importantly, she appealed to colleagues for help:

“…I can’t implement it on my own, I need your help…” (Elizabeth, Site A, Cycle 1, action-planning Meeting).

By developing understanding of the study’s goals and how she contributed to them, Elizabeth became more overtly aware of how the research team related to and relied on the SILs in cycle 2:

“… the SILs could just fill in the checklist without doing it, there is no way for [the research team] to verify scores….she stated she had only just realised this and that we are interested in the process rather than the actual scores…” (Elizabeth, Site A, Cycle 2, post action-planning meeting with researcher)

In working with the wider research team, Elizabeth started to understand her role as a SIL in different ways: seeing herself more as a ‘change agent’ than as a clinician and relatedly shifting how she engaged in the PDSA process. She focused less on following the checklist and more on engaging the wider clinical workforce in the change initiative (see Table 2).

In Cycle 2 Elizabeth pursued changes to process and practice, strongly guided by local drivers of change agreed by senior management, instead of those driven by the checklist findings. Despite this work, Elizabeth’s records of reflexive monitoring describe how the first time she piloted a change herself, there was a breakdown in communication and therefore a barrier, when a theatre supervisor: “…had forgotten to cascade information…” (Elizabeth, Site A, Cycle 2, SIL weekly implementation process report). Elizabeth’s work in cycle 2 centered on understanding change barriers, re-configuring implementation strategies and the roles of stakeholders in these practice changes, to better embed this change (see Table 2).

Change was achieved by making explicit links to organizational knowledge, including the structures and systems already in place such as existing documentation, established actions, trust policies and external influences such as key performance indicators including Best Practice Tariff (BPT), first to herself and then explaining these to staff). This enabled Elizabeth to align to goals shared by the organization and the checklist during Cycle 3, facilitated by pre-existing networks. Coherence and cognitive participation were features of cycle 3, but there was a noticeable shift towards collective action and reflective monitoring. Elizabeth was seen to role-model practices overtly, reflecting and modifying practice changes to improve implementation. Her work additionally energized change already partly embedded. Elizabeth undertook more networking in Cycle 3 to consolidate change involving others, perhaps to anticipate Cycle 4, where change would need to be maintained without her input.

With Elizabeth fully independent in her SIL role, sense-making activities are almost absent from Cycle 4. She understood what the intervention was and knew how it fitted, or not, within her organization. She attempted to change, or not change, items she saw as contextually relevant, and the work needed by her to make sense of her role diminished. Collective action carried the change forward, reducing the emphasis on role-modelling work, whilst the emphasis moved to handing over embedded change to colleagues:

“…Met with Harriet, who has a chief nurse scholarship for one day a week improving care in Theatre Recovery, she is interested in how to help…” (Elizabeth, Site A, Cycle 4, SIL weekly implementation process report).

Findings in Site B

“Working together to make realistic and achievable goals” (Helen, Site B, Cycle 1 Action-planning Meeting).

For Helen, taking the SIL role in Site B enabled her to work at a higher grade (NHS Band 6) within a linked, but different, ward. As shown in Table 3, Helen’s coherence work, in cycle 1, saw her make sense of the intervention’s individual components (checklist, her job description and PDSA process) then as parts of a larger whole. She used reflexive skills to map her current skills to those required by the role: “…Having some NVQ units to look through for my SIL role was really useful…” (Helen, Site B, Cycle 1, SIL weekly implementation process report). Helen also researched ‘leadership styles’ and how to facilitate open discussions in meetings.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Normalisation process at SITE B.

https://doi.org/10.1371/journal.pone.0279651.t003

Making sense of the PDSA model as part of implementation, Helen undertook extensive work to facilitate the cognitive participation and collective action work of her colleagues in preparing for her first action-planning meeting. She provided Nursing and Midwifery Council (NMC) Continual Professional Development (CPD) documents for attendees to use as evidence of CPD they achieved by attending the meeting;. This work increased the relevance of the action planning activity to nurses and their ongoing registration. Recognizing the multi-disciplinary nature of the intervention, she prioritized specific checklist areas for professions and teams to increase their relevance and reduce work for attendees. Helen encouraged a participatory ethos of input and transparency in her meetings: “[checklist item] what do we think? If [working practices are] not good, say…” (Helen, Site B, Cycle 1, action-planning meeting). The meetings did not produce formal goals; she created these later, then sought management approval before implementing. Changes to documentation and recording practices featured heavily, enabling existing practice to be recognised in the checklist scoring mechanisms.

Helen worked to maintain relevance and buy-in of teams or individuals in cycle 2, reducing coherence-related work and increasing work related to cognitive participation. In cycle 2, Helen used her labor as an opportunity for mutual exchange:

“…If I support the promotion of MCA [Mental Capacity Act] and support through the CQC [Care Quality Commission] visit I will hopefully get support when I want things for PERFECT-ER …” (Helen, Site B, Cycle 2, SIL weekly implementation process report).

This reciprocity was unique to this site in featuring strongly across all four cycles (see Table 3).

By tacitly role-modelling practice and documentary changes Helen’s implementation role was key but led to collective action being centered on her own work across cycles 1 and 2. Helen recognized the difficulties with this approach during cycle 3 but continued to use available opportunities to role-model change and to collaborate with others to facilitate implementation. She attempted to embed new practices within the work done by HCAs ensuring that those requiring new skills and competencies became formally recognized in the organization. Producing such recognition within the organization, helped ensure change continued.

Helen brought energy to the implementation, in Cycle 3, by engaging and activating others, for example, arranging and meeting with colleagues engaged in improving care for people living with dementia, to ensure continued investment once her role had stopped [32]. By considering which colleagues she could handover certain changes to, Helen was able to begin to step away from her implementer actions (doing collective action) encouraging others to take these initiatives forward, so securing continued investment and sustaining change. In cycle 4 Helen continued collective action, remaining visible as a ‘doer’, but finalised handover arrangements for changes previously dependent on her efforts.

Findings in Site C

“She felt most of the scores were either 100% or 0% and that some of the 0%s were actually being done, but not recorded…” (Karen, Site C, Cycle 1, pre- action planning meeting briefing with researcher).

Karen had not been a ward-based healthcare professional prior to becoming the SIL at Site C thus, she undertook sense-making to orientate herself to her new role (see Table 4). Coherence work was needed to understand the explicit and implicit rules, structures and ethos of the new context. Karen actively framed the intervention, and her role within it which she saw as a way to: “…highlight all the positive work being done on the ward… [and that]… she would become a voice for patients with dementia across the entire hospital…” (Karen, Site C, Cycle 1, action-planning meeting). As an ‘outsider’ Karen positioned herself as ‘onside’ and thus insider [33]. However, framing the purpose of the checklist, as highlighting “the positive work on the ward” posed a barrier to implementing change when it threatened the ‘virtuousness’ of current practice. She responded to this challenge by asking for changes to the checklist, rather than pushing for changes in practice, to restore the ‘virtue’ of current practice. Karen frequently brought checklist items back to the research team to be altered, illustrating her inability to move beyond coherence as she struggled to internalise the values and benefits of new practices. Instead, she focused on a role as negotiator between Site C and the research team across cycles 1–3.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Normalisation process at SITE C.

https://doi.org/10.1371/journal.pone.0279651.t004

“…We scored 100% [on checklist item] initially, but [research team] have changed the wording… and we score 0%… we will always score 0% and this reflects badly on us…I will take this back [to the research team]… as I’m not happy with this as it is not fair…” (Karen, Site C, Cycle 1, action-planning meeting).

This need to recognize and prioritize current practice rather than trying to modify practice meant that her reconfiguring work focused on trying to change the checklist rather than the related practices.

Implementation can be particularly undermined if those expected to deliver change actually resist it. Karen contested items in particular ways: “…they [Karen and Mohamed] were certain that they did this task, just needed to negotiate with [research team] to tick the box…” (Site C, Cycle 2, action-planning meeting). As a small group they could not make sense of the change, because they could not connect with and thus comprehend the wider systems. Karen was supported by the Site’s dementia team, who assisted with sense-making thereby helping to move (who or what?) towards cognitive participation and collective action by being supportive about elements of the checklist contingent to their remit:

“…[checklist item] was discussed around the table, Mohamed [PPL] wanted to know if it was really needed. Both dementia workers were very supportive…and felt there were lots of dementia-related issues….that needed recognizing. It was agreed that Karen would work with the dementia workers to incorporate a section in the existing policy…” (Site C, Cycle 1, action-planning meeting)

Over successive cycles, other peers became involved, bringing their wider specialist knowledge assisting with sense-making. This redirected the sense-making, providing new energy, facilitating Karen and Mohamed to move from recursive sense-making into cognitive participation and collective action.

In a further example, during the Cycle 3 action-planning meeting, Mohamed questioned whether particular sets of care practices would be a good thing for patients, even with additional resources. Supporters of specific items (a Dementia Worker and a Senior Physiotherapist) repeatedly advocated their legitimacy and importantly began to suggest solutions. The cognitive participation work in this meeting, enabled Mohamed to recognize the importance of getting: “…a new female manager…” on board (Site C, Cycle 2, action-planning meeting). These actions and suggestions for initiating change emerged from others suggesting that even if advocates struggle to see value in proposed changes, peers may show a way to begin to progress changes if they believe it is right for them to be involved. As Karen suggested, “… the new senior physio is really helpful and thinks outside the box…” (Cycle 3, SIL teleconference meeting notes).

Cycle 4 was a winding-down phase at Site C. Karen saw her changes as already handed over and being implemented by others. She stated that she had not been a frontline catalyst for change but had worked to change paperwork with key individuals. She perceived little implementation work was needed in cycle 4 (see Table 4).

Implementing change: Learning across place and time

We have used NPT constructs to analyse ‘novel’ data and present underlying mechanisms across cases and to provide insights into changes over time.

Coherence: An intervention legacy

Coherence work at the beginning of implementation featured strongly in these data and appeared necessary when introducing complex change(s), producing a legacy of shared understanding. At Site A, Elizabeth’s prior contextual knowledge enabled her to set up documentary changes ahead of her first action-planning meeting which, in turn, reduced cognitive participation work later required by others. In comparison, at site C, the change agents (Karen and Mohamed) lacked familiarity with the detailed context and engaged more with sense-making work across cycles 1–4. At site B, Helen’s coherence work had similarities to Elizabeth’s, but she undertook more sense-making work in relation to her role to locate its, and thus her, place within the intervention.

Elizabeth’s early coherence work reduced opportunities for colleagues to input here, whereas at site C, Karen’s lack of familiarity with the context, and thus contextual knowledge, indirectly and perhaps paradoxically, encouraged others (physiotherapists and dementia workers) to become involved to assist her sense-making. At site B, Helen’s coherence work in cycles 1 and 2 encouraged colleagues to engage in sense-making work, empowering them to suggest their own change solutions. In these cycles, cognitive participation and collective action focused on enrolling others to invest in the intervention. However, how Helen’s site B colleagues worked together to make-sense of the intervention (communal specification) and the specific implications for their own roles within it (individual specification), resulted in solutions involving Helen directly as the doer. In cycles 3 and 4 this resulted in Helen having to work to reconfigure, re-initiate and rework embedded changes to move the emphasis away from her.

Normalizing processes to embed complex interventions work best when following a clear pathway to a coherent shared goal. Choosing the path, and retaining responsibility for keeping to it, may require less work if those making the choices do not consult widely. In Site A this meant that over time coherence requirements reduced, but the longevity of this goal, and with it the prolonged commitment of others, could prove challenging when others were not collectively invested. Site A’s implementation relied on overt shared goals between the checklist and the organization but, this mechanism of normalization may not engender the continued investment of colleagues.

Cognitive participation: The costs of widening enrolment

In sites A and B, Elizabeth and Helen held two action-planning meetings per cycle to capture as many colleagues as possible. In site C, Karen held two meetings, however, one of these was a pre-meeting with Mohamed (PPL) where they pre-decided actions. In sites A and C, Elizabeth and Karen used meetings to cascade information about actions they had created. This reduced the requirement for attendees to perform cognitive participation work. Whilst some discussion regarding the feasibility of change occurred, attendees needed to perform little work. In Site B Helen’s work facilitated wider enrolment but at a cost to herself. Helen’s approach to action-planning meetings required more work from her and involvement from attendees. Despite this Helen:

“…was disappointed no goals had been made, but she was happy everyone was giving ideas…She…[will]…look at the ideas put forward and run them past the ward manager before implementing them through informal chats…”

(Helen, Site B, Cycle 1, post action-planning meeting with researcher)

The legacy of cognitive participation in all sites is clear. At Sites A and C people knew what was required of them when they left the action-planning meetings. Cognitive participation quickly gave way to enacting (or not) changes in their practice. Site B staff made greater investment, engaging in more cognitive participation during action-planning meetings. This had to be translated into action plans and cascaded outside meetings by Helen using a combination of word-of-mouth and visible role-modeling of change solutions. This required more work from Helen and highlights that engaging participants widens the number of perspectives, and so increases the work of generating convergence through collective reflexive monitoring.

Collective action: Implementing differently

The SILs engaged with challenges that were complex and variable, set against a backdrop of frequent staff change (clinical rotations) and shortages with or without bank or agency staff replacements. Each SIL pursued change differently. Karen focused on documentary change to embed ward-based change and evidenced this from patients’ notes. This reliance on normalizing change through documentation, and having less physical presence in the clinical setting, meant that implementation suffered: “…they could have jellyfish on their feet for all I know…” (Karen, Site C, Cycle 3). In other sites being physically present in the ward assisted implementation. Helen role-modelled practice, resulting in collective action relying almost entirely on her actions. Only in cycle 3 did Helen activate others, which then brought new energy to implementation.

Prior to their appointments as SILs change agents came from different work environments, (see Table 1). This influenced how they variously implemented the intervention, having different experiences and networks to draw on. The SILs in sites A and B were ward-based with existing connections to ward-based networks. In site C the SIL was not ward-based prior to commencing the role, meaning connections with formal and informal networks required more work. This affected some ward-based change, which may have been facilitated by established connections with ward networks. Such networks, in every workplace, require individuals to undertake relational or contextual integration work. Reflecting on her lack of integration with ward personnel, Karen stated: “…if low scoring items on the checklist had been things like bowels [a ward-based change] perhaps I would have been more embedded [in the ward]…” (Karen, Site C, Cycle 3, action-planning meeting). Our data suggest generating and facilitating networks, as part of collective action work, enables implementation to become more energized and embedded.

Reflexive monitoring: Enhancing or inhibiting implementation

Appraisal work may enhance or inhibit implementation across time. When implementation stalled in a site, rotation of new staff offered opportunities for introducing fresh insights and understanding re-energizing impetus. However, frequent movement of staff provided ongoing implementation challenges: “…she might not follow some of the things we have started…” (Helen, Site B, Cycle 4, SIL weekly implementation process report). Continuation of implicit informal working practices without tangible consequences were fragile compared to practice supported by management and trust policies with formal consequences.

External influences mitigate the fragility of implementation. For example, practice required by the National Hip Fracture Database (NHFD) BPT has a financial reward based on performance. Practice required by this is not threatened by the need for appraisal work, stimulated by frequent staff changes, because of the tangible nature of its consequences. This is reinforced by the formal collaboratives, put in place by trusts, to ensure such financial reward. In contrast the informal collaborative nature of the PERFECT-ER intervention implementation entailed frequent and ongoing appraisal work constant changes in personnel and a continuing need to undertake repetitive reflexive monitoring work.

Sense making, implementing and normalizing change using a PDSA process implicitly required practitioners to be reflective. Reflection does not fit with ‘delivering more for less’ discourses and as such, its legitimacy and value over more performative/presentational work with instantly tangible/visible outcomes is diminished. Reflection is hidden and therefore ‘invisible work’ [34].

Wider implications for implementation

Organisational clinical specialties, and the rotational nature of staffing, create barriers that limit the sustainability of change in implementing new ways of working in acute settings [35, 36]. Discourses which emphasise individual accountability before collegiality increase these barriers by discouraging shared sense-making and shared reflection to inform collective action [37, 38]. ‘New professionalism’, in which repertoires of ‘productivity’ or ‘deliver more for less’ condition individualised professional duties, find traction in busy clinical environments which rely on rationalisation to encourage efficiency but discourage coordinated professionally-informed actions [39]. Thus, when trying to deliver sustainable change, (the) formal organisationally-driven collaboratives can enhance implementation-promoting collective actions which align with rationalised healthcare systems. However, this comes at the cost of reducing coherence, cognitive participation, and reflexive monitoring work.

In the acute setting, a short-term ‘ends justifying means’ approach appeals; this is enhanced by material incentives where organisations/institutions reward conformity (for example, NHFD BPT). However, such approaches may ensure more short-term superficial implementation, which may not change individual’s underlying preferences over the long-term. Relying on SILs as implementors requires considering their skills, their experience, and their knowledge of the working setting, all of which will frame their approaches to implementation. To attend to the potential consequences researchers should review both key implementors’ training and their relationships with research and clinical teams.

Restricting individuals’ autonomy may be necessary when an organization’s resources are constrained by hierarchical structures [40], but this can pose a further barrier to change.

If staff autonomy is restricted in the medium to long-term, this affects professional socialisation, a potent social influencing and sense-making mechanism, and so would limit normalisation [41]. The paradox, therefore, for implementing change is to focus on cultures and mind sets relevant to the context, but to also to recognise that bringing people together from different disciplines to be involved in change, will not provide a ready-made uniform culture form to underpin an informal collective’s (coherent) sense-making, or indeed to become a cohesive group. Nor will this readily provide an informal set of working practices required to get things done (collective action). Bringing together different groups and disciplines might therefore lead to more piecemeal and less embedded change. Thus, agents must work to actively cultivate, review, and maintain formal and informal networks, continually reinvesting in these, over time, to sustain engagement [42]. Using NPT to examine change across sites and over time using ‘novel’ data to make has made visible the professional and economic investment needed to maintain change and the context-based facilitative factors. Using such data provided temporality not normally seen in focus group or interview data capturing the ongoing process of implementation.

Limitations

We sought to implement and refine a new multicompetent intervention in acute hospitals. We relied heavily on SIL generated data without triangulating from other data sources such as other staff in the settings, organizational leads who may have offered alternative perspectives and views and their internal networks. Different change agents had similar disciplinary backgrounds, but their knowledge of ward environments still proved to differ from each other. We have not therefore investigated how other staff disciplines may have influenced implementation.

Conclusions

We have presented examples of implementation and normalization over time (4 cycles of change) and place (3 acute hospital wards). Despite challenges intrinsic to implementing a novel intervention, each change agent adopted a method of implementing and normalizing change in their ward which could be seen to have contextual fit. Staff collectively activated and operationalized those features that aligned to what motivated in that context. Nonetheless this did not lead to the content and legitimacy of PERFECT-ER being evenly adopted as agents continued to encounter resistance(s) at times and in place. Successful implementation and subsequent normalization were seen to require alignment between perceived values, fit and wider contextual goals. This was important to revisit to ensure sustainability, particularly given the context of constant change which can fragment normalization in both organization and staffing. This meant the role of the change agent (SIL) went beyond successfully introducing the initiative as they continued to sustain the recursive two-way sense-making process required by organizations so as to recognize and respond to staff changes and to engage new energy from incoming staff.

This study reveals how recursive collective sense-making can be. Individuals may not be able to make sense of what they are implementing until different perspectives are brought in. Sense-making moves from individual to collective but also from collective to individual. Fragmented collective insight(s) are brought together by key individuals (in this case the SIL) and then fed back to wider stakeholder groups to foster new collective sense-making and impetus, thereby overcoming impasses in implementation and normalization and working towards sustaining change.

Implementation can be successful, but this is often at some considerable cost to those trying to implement change. Thus implementation–if it happens, is consequently fragile as the present NHS context does not prioritise service improvement and the efforts required for this task thus appears contextually illegitimate.