Choice of normalisation process theory for process evaluation

The development of this complex intervention was informed by a pragmatic clinical case note review with input from two specialties–palliative care and geriatric medicine [3]. There is now increasing evidence that the successful development and implementation of complex interventions benefit from the contribution of theoretical frameworks [31–33]. As part of the feasibility cRCT of the AMBER care bundle [22], we conducted a process evaluation to understand how the intervention was operationalised, and what modifications and refinements were needed to optimise its use in acute hospital settings. To achieve this, we made use of Normalisation Process Theory (NPT) [34], a socio-behavioural theory. NPT was chosen as it focuses on the ‘social organisation of the work (implementation), of making practices routine elements of everyday life (embedding) and of sustaining embedded practices in their social contexts (integration)’ [34] (p 538).

NPT provides a set of tools to identify and explain the social processes through which new or modified practices of thinking, enacting and organising work are operationalised in institutional settings: in this case, hospitals [21]. Moreover, it sets out a three-stage model of implementation, embedding and integration and is organised around several important questions: (i) What factors promote or inhibit the routine incorporation of the intervention in practice? (ii) What factors promote or inhibit the implementation, embedding and integration of the intervention? (iii) What factors promote or inhibit the mobilisation of structural and cognitive resources for the implementation of the intervention? [35]

The theory identifies four essential determinants of embedding or ‘normalising’ complex interventions into common practice. These are (i) ‘coherence’–the extent to which an intervention is understood as being meaningful, achievable and valuable; (ii) ‘cognitive participation’–the engagement of individuals (in this case, health professionals) necessary to deliver the intervention; (iii) ‘collective action’–the work that brings the intervention into use; and (iv) ‘reflexive monitoring’–the on-going process of adjusting the intervention to keep it in place [22]. These components are considered to be dynamic and interact within the wider context of the intervention, such as existing organisational structures and procedures [36]. Importantly, they are in keeping with the UK’s Medical Research Council (MRC) guidance on the development and evaluation of complex interventions [37] and the Methods of Researching End-of-life Care (MORECare) statement [15], which both stress the importance of theory in understanding what makes interventions effective.

Aim and objectives

In this paper, we examine the use of NPT to determine if, and in what ways, the AMBER care bundle can be successfully embedded or ‘normalised’ into acute hospital care practice to support patients whose situations are clinically uncertain, and their families. More specifically, the paper aimed to integrate our data across sources under NPT constructs to (i) understand individual and contextual facilitators and barriers surrounding the implementation of the AMBER care bundle and (ii) identify strategies to strengthen facilitators and mitigate barriers, informing the optimisation of the intervention and its future sustainability in acute hospital clinical care.

Design

We made use of a multi-method design [38] within the wider mixed-method feasibility cRCT trial [22]. Data were collected in parallel and then analysed and integrated using NPT. The data collection approaches included (i) health professional focus groups, (ii) interviews with patients and/or relatives, and (iii) multi-disciplinary team meetings’ non-participant observations. The quantitative component involved a detailed examination of patients’ clinical notes. The study was registered in a freely accessible clinical trial registry (ISRCTN36040085).

Study setting

The feasibility cRCT took place across purposefully selected general medical wards located in clusters. For the purposes of this study, the clusters were represented by four district general hospital (DGHs) in England. Implementation of the AMBER care bundle were limited to two of the four DGHs. Selection of study wards at each site was informed by heat maps that provided contextual information at a ward level on the number of deaths during and up to 100 days after admission. Wards with the highest number of deaths per year were considered to be suitable for the study [22, 29]. In this paper, we focus on the two intervention sites, where the AMBER care bundle was implemented and delivered and who provided data for the process evaluation reported in this paper Table 1. Full details of how the sites were selected are reported elsewhere [22].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Description of study sites.

https://doi.org/10.1371/journal.pone.0239181.t001

Implementation of the AMBER care bundle

A nurse facilitator supported the implementation of the AMBER care bundle across the two intervention sites for a period of two months. This involved: familiarisation with the ward, introducing the intervention to health professionals and training them on its use, supporting them in the practice of using the intervention, and observing how they used it in practice. Full details of the implementation of the intervention across study sites is reported in detail elsewhere [22, 29].

Sampling

Data collection

Questions asked during (i) focus groups with ward staff, (ii) semi-structured interviews with patients and/or relatives; (iii) issues noted during the non-participant observations of multi-disciplinary team meetings; and, (iv) the review of patient participants’ clinical notes, were informed by our patient and public involvement (PPI) representatives, and were aligned to NPT.

Analysis

Interviews with patients and relatives, focus groups with health professionals, and non-participant observations of multi-disciplinary team meetings.

Qualitative data were analysed using the Framework approach and thematic analysis [41] facilitated by NVIVO 10. Members of the research team who led on conducting the interviews, focus groups and observations (JK, EY and HJ) also led on the analysis. They familiarised themselves with the raw data and discussed their impressions of the dataset. The NPT constructs–‘coherence’, ‘cognitive participation’, ‘collective action’ and ‘reflexive monitoring’–provided a thematic framework Table 2. We took a robust approach to analysis: all interviews and focus groups were double coded in NVIVO 10 by two researchers (EY and JK, EY and HJ, or JK and HJ) independently followed by comparing results and discussion within each researcher pair to ensure uniformity of coding. We also hosted ‘data workshops’ where the researchers coded a sample of transcripts together with our patient and public involvement (PPI) members to minimise bias in interpretation and the validity of findings. Once transcripts were coded, data was exported from NVIVO 10 and were charted and mapped using Excel^®. Non-confirmatory data were also explored and consideration was made about their sources to avoid making unwarranted claims about patterns in the data [43].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. NPT constructs relevant to the AMBER care bundle.

https://doi.org/10.1371/journal.pone.0239181.t002

Data across all sources were then discussed further by the researchers and triangulated using the NPT constructs to understand the operation of the AMBER care bundle on each ward. At this stage, researchers identified areas of confirmation and contradictions across sources which were used to greater researchers understanding of the operation of the AMBER care bundle across health professionals, and the contributing facilitators and barriers involved in the sustainable use of the intervention.

Research governance and ethical approval

Ethical approval was obtained from the National Research Ethics Committee—Camden and King’s Cross (20.12.2016, REC Reference: 16/LO/2010) and Health Research Authority (25.01.2017). Local research governance approvals were obtained from participating hospitals.

Coherence–making sense of, and finding meaning in, the AMBER care bundle

Coherence represented the process through which ward staff shared a common and valid interpretation of the purpose and value of the AMBER care bundle. Overall, ward staff were observed as having a good practical understanding of the intervention and its constituent components. For example, during the non-participant observation of a morning handover at Site 1, a range of health professionals were noted as being confident and clearly explaining the intervention to a new consultant on the ward.

There was broad agreement from health professionals that the intervention represented a positive shift in the emphasis of care for a patient group who were previously overlooked in clinical practice. Ward staff recognised that the intervention prompted them to recognise and prioritise patients whose situations were clinically uncertain and further, engage in important discussions with them regarding preferences for care, escalation decisions and medical treatment. Many noted the value of this for ensuring that patient and family preferences were discussed, captured and communicated with those involved in the patients’ care in a timely manner.

The AMBER care bundle was also perceived to be valuable in supporting some staff to provide care for this patient group. At Site 2, the ward manager highlighted the value of the intervention for junior staff. First, in increasing their understanding of clinical uncertainty, deterioration, and end of life and secondly, their increased confidence to engage with this patient population by using the intervention as a platform to broach such topics.

… But now with the ‘AMBER’ I think they can talk, and they will feel more confident to talk with relatives.

(Site2-014, female, Ward Manager)

Whilst most health professionals suggested that the AMBER care bundle represented a fundamental change in care, a small number did not believe it differed noticeably from their existing practice. These individuals did, however, note that the intervention acted as a means to formalise their current practice, a view typified by the following junior doctor:

Speaking for myself, if I have someone who I’m worried and has the potential of deteriorating, I would always within my best capacity to try and inform the family about their situation. I don’t think it changed our practice. The only thing we’re doing is just formally documenting it. We were doing everything we could, uhm yeah, even prior to AMBER bundle.

(Site2-005, Female, Senior House Office)

Despite health professionals holding a coherent view of the value of the AMBER care bundle, a lack of clarity surrounding the intervention’s eligibility criteria resulted in a varied understanding of what patients were most appropriate for the AMBER care bundle. In practice, the clinical team were frequently observed making judgements on patients’ suitability to be supported by the intervention, based on the patients’ level of co-morbidity, frailty, disease progression, the likelihood of responding to treatment and medication, or their ‘ceiling of care escalation’, that represented proxies for ‘clinical uncertainty’. They rarely referred to a patient’s ‘risk of dying’ during the episode of care [the admission] to inform decisions. This was further exemplified by the responses received by health professionals during focus groups who when asked to describe the patients who were suitable to be supported by the intervention used descriptors such as ‘those who are aged and frail’, or ‘those with an unpredictable recovery’.

One junior doctor on Site 2, explained that they had focused on identifying patients in clinically uncertain situations rather than those at risk of dying, due to the latter requiring them to prognosticate, which they did not feel skilled or confident to do. This posed fundamental challenges and often made this aspect of the eligibility criteria impossible to operationalise coherently. Further, this reluctance to acknowledge a patient’s ‘risk of dying’ was noted by patients and their family, who were often more aware of their own, or their loved one’s likelihood of death.

I remember having a conversation with the doctor and saying, “Do you really actually think he’s going to be discharged out of here? Because he looks like he’s a dying man to me.” The doctor just said to me “You have to be optimistic”, and I just said, “Optimistic or realistic?” you know?

(Site1-017, Carer)

Cognitive participation—Commitment and engagement with the AMBER care bundle

Cognitive participation represented the extent to which key stakeholders (health professionals) were adequately motivated to incorporate this complex intervention into their practice and how well it fitted in with existing approaches. All staff were expected to engage in the active identification of patients whose situations were clinically uncertain, discuss plans for treatment and care with patients and their family, and document these in patients’ clinical notes.

Across sites, staff participating in the focus groups were unanimous that the role of the nurse facilitator was critical in successfully engaging with and operationalising the intervention. She encouraged them to appraise which patients might be suitable for the AMBER care bundle, as well as, importantly, prepare and reflect on conversations with patients and their families which were often difficult and emotionally demanding. Observations of multi-disciplinary team meetings supported these views. At numerous points, the nurse facilitator was observed encouraging health professionals to complete AMBER care bundle components, particularly in identifying instances of clinical uncertainty, ensuring important discussions took place with patients and families and reminding staff of the requirement to document these circumstances.

He’s ‘AMBER’, so while writing the discharge letter, we should remember to note the things discussed and escalation decisions.

(FIELD NOTE: Site2-011, Female, Nurse Facilitator: observed during Handover meeting)

Importantly, because of their pressured workloads, ward staff did not believe they had additional capacity to ‘champion’ the intervention in the ward within their roles. During focus groups, they questioned their ability to engage in the delivery of the intervention without the continued dedicated support from the nurse facilitator. Specifically, they highlighted the challenges and time required to train new staff about the intervention, a situation amplified by increasing levels of staff turnover.

Staff were fearful of potential negative consequences which may come from the use of the intervention without a dedicated ‘champion’ to facilitate engagement. At the very least, there was concern that they would overlook patients who might benefit from the intervention and, more seriously, a wariness of engaging in potentially emotionally laborious conversations with patients and or their relatives.

I think the nurse facilitator was excellent actually in helping us implement and ‘do the AMBER’. Since she’s gone, I think it dropped off a little bit.

(Site1-022, Female, Ward Manager)

Collective action—Work required to make the AMBER care bundle function

Collective action represented the notion that ward staff performed actions or tasks based on principles outlined by the intervention. The data from across the different components of the study shed light on the resources and procedures associated with its integration into routine practice. Since the general principle of delivering patient-centred care was shared across staff, for some, the intervention was not perceived to represent a radical departure from their current way of delivering care for some individuals. This consequently meant that the work associated with the collective delivery of the intervention was generally well accepted into daily clinical practice.

Nevertheless, formally operationalising the daily clinical activities associated with the intervention provided a welcomed opportunity for other ward staff to be involved in discussions about the decision-making process associated with the delivery of care for this patient group. Instances of teamwork and shared decision making were observed during the handover meetings. At Site 2, for instance, a range of health professionals contributed to discussions about patients’ suitability for the AMBER care bundle. In one focus group, a consultant also highlighted that whilst it was doctors who were typically perceived as being pivotal in patients’ care, the views of other professions increasingly began to contribute to patient-centred decisions and their medical plans. In this respect, the presence of the intervention enabled and empowered those who had previously not been called on for their views, to actively share them. A consultant on Site 1 emphasised the shift from clinical decisions being made independently to decisions made as an MDT, with health care assistants who often have more insight into the day-to-day care of the patient, and allied health professions now being involved in decision making.

Perspectives from some patients and relatives substantiated health professionals’ views, highlighting teamwork across professionals and specialties during their care, or the care of their family member:

100% work together well.

(Site2-009, Male, Patient)

The OTs especially have been brilliant there … and yeah they seem to, they seem to all know what’s not and what’s going on.

(Site2-016, Female, Carer)

The case note review of patient participant notes provided further insight into ward staffs’ involvement in patient-centred decisions and medical plans Fig 1. In Site 1, 90% of patient participants’ notes detailed discussion and agreement of patient medical plans with ward staff, and in Site 2, this was noted in 67%, demonstrating documented involvement of other health professionals.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Documentation of AMBER care bundle components in patient participants’ clinical notes.

https://doi.org/10.1371/journal.pone.0239181.g001

Further analysis of the patients’ clinical notes provided additional insight into the collective action taken to complete the required documentation associated with the delivery of the intervention. Overall documentation of medical plans and escalation plans for patients across sites were high (medical plans 90% and 78% and escalation plans 80% and 78%, for Site 1 and 2 respectively).

This high level of compliance may be due to health professionals perceiving the documentation required as part of the AMBER care bundle as valuable, simple, and easy to complete.

Despite this, there were, however, collective action-related concerns with some of the intervention procedures, particularly those that relied on timely discussions between health professionals and patients and/or their relatives. The combined findings from the focus groups, interviews with patients and relatives, and review of clinical notes identified that the completion of initial conversations and follow-ups with patients or their family were often delayed. In focus groups, health professionals reflected that this was often because many of the patients suitable for the intervention were elderly, confused and lacked mental capacity. Consequently, engaging in conversations with these patients about their situation, and making treatment and care plans, necessitated the presence of family.

Relatives, however, were not always available to have important discussions in a timely manner. For instance, some were unable to visit or were only able to visit on weekends or evenings, when health professionals most familiar with the patient’s circumstances were not on the ward. Observations of multi-disciplinary team meetings further evidenced the challenges ward staff reported experiencing in accessing some family members. For instance, when talking about a patient on Site 1, a consultant at the multi-disciplinary team meeting was overheard suggesting a conversation was required with the patient’s wife to discuss the possibility of a referral for a dementia assessment. However, an occupational therapist mentioned that the patient’s wife was herself unwell and currently receiving chemotherapy, so any discussion soon would be unlikely.

A key component and goal of the AMBER care bundle is the requirement for staff to document patients’ clinical situations and medical plans to ensure all staff, including weekend and out of hours staff, are familiar with a patients’ situation. While patient participant clinical notes highlighted that this documentation was occurring in most cases, the detail in these notes often varied from patient to patient, translating to instances of inconsistency in care.

This is exemplified by a daughter of a patient on Site 1 who, when asked if she felt like she was getting consistent information from nurses and doctors, highlighted the importance of all staff being aware of a patient’s clinical situation to ensure they are receiving consistent care.

I think probably some of the nurses, I don’t want to criticise them but I think probably some of the nurses who were maybe, bank nurses and things, who didn’t know her [the patient] so well didn’t quite realise about the delirium and thought it was the dementia.

(Site 1–002, Carer)

Similar experiences were shared by other relatives. Another daughter of a patient on Site 1, highlighted instances where her father had been in receipt of unnecessary tests from some health professionals who did not seem aware of the de-escalation for his clinical situation.

During engagements with patients and relatives, health professionals were encouraged, as part of the intervention, to discuss future plans, including preferred place of care and preferred place of death. Despite this, the clinical notes of patients demonstrated that these preferences were often inadequately recorded, particularly place of death. This was documented in only 30% and 44% of patient clinical notes for Site 1 and Site 2, respectively. Compliance with these components of the intervention appeared to be lower than other care plan elements, such as completion of escalation plans and do not attempt resuscitation (DNAR) discussions. This corresponded with the views of patients and relatives who suggested that, while other areas of care were being discussed (such as treatment), preferred place of care and preferred place of death were often absent from these conversations.

No, we didn’t have any of those conversations in the hospital, and to be honest I haven’t had any of those conversations since he’s gone to the nursing home, which we need to have really to be honest.

(Site 1–017, Carer)

One consultant explained that discussions surrounding the preferred place of care or preferred place of death were difficult for health professionals, due to concerns they might not fulfil patients’ or relatives’ expectations if preferences could not be realised due to over-riding clinical reasons. One consultant explained that for some patients, while the preferred place of death is home in reality, it is unlikely that these patients would be able to be supported at home, due to the services they need and care they would require. There was therefore reluctance to engage in these conversations with patients in fear that they may be disappointed if preferred place of care or death could not be achieved.

Findings also highlighted that communication with patients and families was often variable due to workload pressures. During conversations, staff felt rushed and unable to provide adequate time to have these important discussions with patients and their relatives. This proved to be frustrating and health professionals believed that, whilst the intentions of the intervention were honourable, the fidelity of its delivery was often compromised. This was typified in the following comment from a registrar:

Sometimes our communication is brief, just because of the time pressures we are in nowadays. Even when I have the conversation, I sometimes feel that if I spent 10 minutes more or 15 minutes more with a patient, I’d probably explain a lot better.

(Site2-002, Male, Registrar)

Reflexive monitoring—Opportunities to appraise the AMBER care bundle

Reflexive monitoring refers to the appraisal process which health professionals, formally and informally, undertook to assess and understand the ways that the AMBER care bundle affected themselves, patients and/or families and others around them. Opportunities for formal reflection and monitoring were not integrated into the intervention. However, we did identify, through observations and focus groups, attempts by health professionals to locally modify or reconstruct the AMBER care bundle and its delivery, to successfully enable implementation within their acute hospital setting. This included adapting the frequency of reviewing patients’ situation of clinical uncertainty and removal of the requirement to place a yellow ‘A’ sticker in patients’ notes Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Modifications to the AMBER care bundle component.

https://doi.org/10.1371/journal.pone.0239181.t003

Optimising implementation, integration and sustainability of the AMBER care bundle.

Fig 2 presents a model of the facilitators and barriers of the AMBER care bundle alongside modifications that must be considered to enable the normalisation of the intervention within the acute hospital setting. In response to these barriers, improvement strategies are identified that are likely to contribute to improving aspects of cognitive participation and collective action to sustain practices associated with the delivery of this complex intervention [14, 44].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Facilitators, barriers and strategies for normalisation of the AMBER care bundle in acute hospital settings.

https://doi.org/10.1371/journal.pone.0239181.g002

Aspects that aided normalisation of the AMBER care bundle

The intervention was used by health professionals to prompt them to actively consider the care and management of a patient group which were not previously prioritised in their clinical practice. The intervention itself acted as a platform to broach the issue of clinical uncertainty within the acute hospital setting, helping to shift the focus of care provided. This symbolic use of the AMBER care bundle is in line with previous qualitative explorations which identified health professionals use of interventions to prompt changing the focus of care [24, 46]. For some health professionals, the AMBER care bundle was perceived to differ little from their current ways of working. Considering the growing number of end of life care-related support aids or tools, for example ReSPECT [10], a lack of differentiation may be present and problematic, with some staff resisting interventions they believe may replicate existing practices and reflect an unnecessary administrative burden [14]. Yet most health professionals perceived little administrative burden associated with the AMBER care bundle, with the required documentation considered simple, clear and easy to complete. Further, while health professionals recognised the value of the AMBER care bundle for improving their own processes (simple and clear documentation, the involvement of all staff in decision making), they were also clear on the value and benefit of the intervention for patients and their families, including empowering patients to share their preferences earlier, and involving patients and their family in critical decisions. This is a welcomed shift away from an overemphasis on processes to a greater focus on the ultimate goals of intervention delivery: improved care for patients and their families [46].

Critical to initial and sustained engagement with the intervention was the provision of a ‘champion’ (nurse facilitator) who ensured that health professionals were prompted to consider all components of the intervention and were supported throughout delivery. The importance of a ‘champion’ has been previously emphasised for the successful implementation of integrated end of life pathways within organisations [10, 47]. For the AMBER care bundle, the ‘champion’ promoted active engagement, trained new staff members and provided emotional support to health professionals delivering the intervention.

Aspects that were disadvantageous to the normalisation of the AMBER care bundle

Staff shared views on their concerns associated with identifying patients suitable for the AMBER care bundle which currently require staff to judge a patient’s ‘risk of dying during their episode of care’ alongside their clinical uncertainty and deterioration. In practice, health professionals made decisions about who was suitable for the intervention based more broadly on their capacity to benefit from it, rather than specifically their ‘risk of dying’. This highlights health professionals unwillingness to make judgements based solely on prognostication and is in keeping with current evidence which notes the difficulty some health professionals have recognising impending death [48]. Further, research suggests these judgements are highly subjective and are frequently inaccurate, which raises important issues for the operationalisation of the intervention [48]. Solutions to this issue were present during the development of ReSPECT [10]. During the development phase of the intervention designers incorporated cognitive interviewing to understand exactly how terms were interpreted across different professional groups and levels of seniority in the field, prior to its wider-scale implementation [10]. The results of this work contribute to ensuring intervention fidelity and also offer insights into where a more nuanced, flexible interpretation of the intervention eligibility criteria may be required. In its current form, we provide evidence that the eligibility criteria of the AMBER care bundle warrants serious consideration since it limits opportunities to potentially escalate care for those patients who might benefit from critical or high dependency care settings [49], as was originally intended by this intervention [3]. This has potential implications for patient experience and outcomes.

Inadequate skills and confidence in communication also led to issues in the delivery of the intervention. Variation in staff skills meant that the delivery of difficult and daily follow up conversations with patients and their family were frequently allocated to specific health professionals to complete (those with greater experience and confidence in communicating), creating workload pressures and delays in delivery. Further, the emotive context surrounding difficult conversations relating to clinical uncertainty has the potential to impact on health professionals’ well-being, particularly among some junior doctors [50]. This is pertinent in light of the ongoing global COVID-19 pandemic [51] where a growing number of relatively inexperienced junior doctors are being called upon to engage with profoundly unwell patients whose situations may be clinically uncertain, in addition to their very distressed families. Caring for patients in this situation is especially challenging, requiring additional skills, excellent communication, and importantly, systems of support for those delivering care [16, 52]. The absence of such systems may contribute to health professionals experiencing what has been referred to as ‘moral injury’ [53], representing deviations or transgressions from health professionals’ moral beliefs and expectations that are witnessed, perpetrated, or allowed by that individual. When not acknowledged and supported, this can lead to stress, depression, anxiety and post-traumatic distress.

Contextual issues, particularly workload pressures and time constraints were also cited as a significant barrier to the delivery of intervention components. Similar to previous studies exploring the implementation of complex interventions into clinical practice [47, 54, 55] frequent staff-turnover and competing priorities led to challenges and meant that health professionals felt unable to ‘champion’ the intervention themselves, raising concerns about the sustainability of the intervention.

Notably, many of the patients who the health professionals believed suitable to be supported by the AMBER care bundle often lacked mental capacity. This was associated with additional time needed to contact the family to be involved in decision-making, creating further workload pressures, and often resulting in delays in important discussions. Having a clear understanding of the population the intervention intends to serve is critical to minimise harmful outcomes and is emphasised in the Medical Research Council and MORECare guidance [15, 37, 56].

Supporting normalisation of the AMBER care bundle in acute hospital settings

To support the normalisation of the AMBER care bundle in acute hospital wards, important considerations regarding the skills, knowledge and attitudes of those delivering the intervention, and an understanding of the context in which it exists, is required. First, staff need to have a coherent view of who is suitable for the AMBER care bundle. It is therefore imperative that the AMBER care bundle eligibility criteria focus solely on a patient’s ‘clinical uncertainty’ and their clinical needs. Criteria that avoid the need for prognostication may ensure patients who would benefit from being supported by the intervention can be quickly, and more efficiently, identified by health professionals [16].

Second, staff need to be equipped with adequate communication skills to engage clearly and compassionately with patients and their family and are allocated protected time to de-brief following difficult encounters. Advanced communication training and integrated opportunities for emotional support therefore need to accompany the AMBER care bundle to ensure staff are skilled to deliver the intervention effectively. This is imperative for nursing and ward staff who are often more present on wards and therefore have potentially more opportunities to communicate with patients and their families. Additionally, previous research highlights the need to ensure this training is continually re-visited to further develop and strengthen staff knowledge and skills [50, 57, 58].

Last, while patients’ clinical uncertainty may not change on a day-to-day basis, and the clinical team should adapt the frequency of reviewing this to their local needs, it is imperative that brief daily updates for patients and their family are still prioritised. These brief, but frequent updates should aim to provide patients and their family with an update on any medical and/or care changes and provide an opportunity for patients and family to discuss any changes in preferences.

Implications for clinical practice

The AMBER care bundle carries the potential to address a crucial gap in clinical practice and further refinements, as suggested above, would help ensure it is adequately normalised into practice. Patients and their families, who experience uncertainty in light of their conditions, appreciated communication from health professionals and found the opportunity to discuss issues that matter to them valuable. One advantage of the AMBER care bundle is the fact that it relies on the shared human experience of uncertainty. For health professionals, this uncertainty is about the prognosis and likelihood of imminent death or the possible outcomes of the active treatment. Clinical uncertainty can exist for many conditions and has been argued as being an inevitability [59]. While emotive discussions, linked to expressing clinical uncertainty, have generally been seen as the job of designated health professionals, the recent COVID-19 global pandemic [60], has brought with it instances where more health professionals are required to do this work. Specifically, the COVID-19 pandemic has witnessed health professionals dealing with an extraordinary number of cases in acute hospital settings where clinical uncertainty is omnipresent across multiple levels including within the health-system, among professionals, and the patients and their families, for which the course of disease deterioration and potential for recovery is still relatively unknown [51, 61, 62]. As such, with considered modification, the AMBER care bundle has potential to offer health professionals with an approach to better serve patients affected by the COVID-19 and their families.

Strengths and limitations

Our approach used multiple data sources to inform our findings. This represented a practical and feasible way to explore NPT constructs associated with the implementation and operationalisation of the AMBER care bundle in acute hospital care settings, with qualitative components permitting us to explore salient contexts and mechanisms in more detail. Furthermore, the use of four data sources, and data being analysed iteratively by an inter-professional team, including input from our PPI members, should increase the dependability of the study findings and interpretations.

The use of the four NPT constructs as an analytic framework enabled us to provide an understanding of how the AMBER care bundle did, and in many instances could not become normalised within an acute hospital setting. Associated with this, we also identified barriers and facilitators to the future successful integration and implementation of the AMBER care bundle.

This study is also subject to limitations. It represents one component of a wider feasibility cRCT of the AMBER care bundle and was restricted to just two study sites.

Second, we are also mindful of the absence of nursing representation in one of the focus groups, who may have provided additional insights into the use of the intervention in practice from this professional group. Although, several nurses expressed interest, and confirmed their availability beforehand, on the day of the focus group no nurses were able to attend, due to urgent clinical commitments. This highlights the issues faced while researching in a ‘real-world’ context and could be overcome in future studies by considering additional flexibility and resources in the study design to accommodate the unpredictable nature of clinical work. For example, holding two smaller health professional focus groups at each site, or following up individually with health professionals who were unable to join focus groups at the allocated time.

Third, we were not able to conduct direct observations of care due to ‘real-world’ resource and logistical constraints associated with the feasibility study. Direct observations of clinical practice would be possible if research ethical approval for the study included a provision for consent at the level of the unit/cluster level [63].

Last, whilst the model has been informed by NPT constructs, developed as a result of detailed discussions within the research team, the modifications suggested as a result of this study to the AMBER care bundle have yet to be tested in the field.

Future exploration of the AMBER care bundle across other care settings and professional groups will be valuable in providing further understanding of the normalisation of this intervention in practice.