The AMBER care bundle for hospital inpatients with uncertain recovery nearing the end of life: the ImproveCare feasibility cluster RCT

The magnitude of dying in hospital settings

Every year, more than 500,000 people die in the UK,2 and this number is rising. 3 Of these deaths, more than half occur in hospital. 4 However, 69.2% (range 51–84%) of patients and their families would prefer to die at home. 5 Most deaths are anticipated, with up to 75% of all deaths expected, so there is time for discharge to home or more familiar and preferred surroundings. Major reasons for hospital death are poor communication about declining health between patients and HCPs6 and poor identification and management of patients whose situations are ‘clinically uncertain’. 7

Clinical uncertainty in hospital settings

Clinical uncertainty is not a simple concept and a situation of uncertainty usually results from several inter-related factors. Mishel8 was one of the first to develop an overarching theory of uncertainty in an illness, and aimed to explain the underlying processes governing patients’ experiences of uncertainty. Specifically, four concepts contribute to an uncertain state: complexity, unpredictability, ambiguity and lack of information. 8–10 McCormick11 further developed these ideas and described situations of uncertainty in terms of the probability of events occurring, the temporality of events and individuals’ perceptions of their situation.

If clinical uncertainty is not explicitly addressed, there are worse psychological outcomes for patients. 12,13 Evidence suggests that, in the last 30 days of life, the combination of deteriorating health and clinical uncertainty are highly distressing for patients in a hospital and their families. 14,15 This distress is amplified when discussions about their situation and preferences for care and location of death are absent; 67–80% of people want to be informed about poor prognosis. 16 However, research shows that discussions about prognosis rarely occur,17 increasing the likelihood of hospital deaths, but also leading to poor satisfaction, mistrust and loss of confidence in HCPs. 18–21 Indeed, complaints about care at the end of life in hospital settings are frequent. 22

Clinical uncertainty also affects clinicians’ confidence and their practice. Clinicians frequently struggle with uncertainty, which can result in overtreatment or overinvestigation,23 increased costs13 and lack of communication with patients about their future. 24,25 Furthermore, clinicians often feel inappropriately trained to deal with uncertainty; only 4 out of 21 UK postgraduate medical training curricula contain detailed recommendations and curriculum goals relevant to dealing with uncertainty. However, when situations of uncertainty are acknowledged and managed alongside high-quality care, particularly at the end of life, collaborative decision-making is possible. 26 This empowers patients and their carers,27–29 and in turn leads to improved outcomes and increased satisfaction with care. 30,31

The potential for better care

Poor hospital care and inadequate communication, which include patient safety and adverse events, have received increasing attention, particularly regarding the frail elderly and the dying. The Francis Report,32 the independent review of the Liverpool Care Pathway for the Dying Patient33 and the Parliamentary and Health Services Ombudsman’s report34 into complaints about end-of-life care all highlight the devastating effect poor communication and lack of honesty can have on patients and their families towards the end of life. Research has also demonstrated that the costs of patient care in the last year of life are high. 35 Yet when specialist palliative care services are available in hospital and community settings, health service costs can be reduced and patient- and family-centred outcomes improved. 36–38

In 2010, a London hospital identified inconsistencies in the quality of care for patients whose situations were clinically uncertain, for those who were deteriorating and especially for those at the end of life. 34 Issues included inadequate decision-making and poor engagement with patients and carers. A potential solution to caring and supporting patients and their relatives in this situation, developed under the umbrella of a bundle (Box 1 explains the concept of a ‘care bundle’), is referred to as the ‘AMBER care bundle’, where AMBER stands for Assessment; Management; Best practice; Engagement; Recovery uncertain.

The AMBER care bundle follows an algorithmic approach to encourage clinical teams to develop and document a clear medical plan, considering anticipated outcomes and resuscitation and escalation status, and to revisit the plan daily. The AMBER care bundle encourages staff, patients, and families to continue with treatment in the hope of a recovery, while talking openly about preferences and priorities should the worst happen. The bundle is modelled on previous empirical work that includes a literature review and examination of clinical records to determine need. 34 However, this bundle was developed to address an identified gap in clinical practice, hence it does not have theoretical underpinnings. Once staff have identified patients who are deteriorating, whose situations are clinically uncertain, with limited reversibility and who are at risk of dying during their current episode of care despite treatment, the multidisciplinary team (MDT) is expected to complete the four tasks within the care bundle. This involves asking:

1. Has a medical plan been documented in the patient records that includes current key issues, anticipated outcomes and their resuscitation status?

2. Has an escalation/de-escalation decision been documented (ward only, high dependency, intensive care)?

3. Has the medical plan been discussed and agreed with nursing staff?

4. Has a patient/carer discussion, or meeting, been held and clearly documented?

With this information, HCPs consider whether or not the patient is still suitable for the intervention, if there had been any medical changes, and if they need to speak with the patient and their relatives daily.

The patient/carer discussion of the AMBER care bundle includes:

• talking to the patient and family to let them know that the clinical team has concerns about their condition, and to establish their preferences and wishes

• deciding together how the patient will be cared for should their condition change (i.e. worsen).

The intended benefits of the AMBER care bundle include:

• increased and improved communication

• enabling/supporting informed and shared decision-making and choice during end-of-life care

• improved patient-/family-centred quality of life by reducing anxiety

• enabling home death if preferred

• supporting HCPs to develop knowledge, skills and confidence in end-of-life care delivery

• reducing unnecessary hospital admissions while improving cost-effectiveness (reducing length of hospital stays) and making more efficient use of health services.

The AMBER care bundle is a complex intervention in that it:

• Comprises multiple components and layers (identification, current and future care planning including escalation and de-escalation decisions, communication delivery, assessment of patient preferences and systematic follow-up, acknowledging dynamic wishes and physical conditions).

• Aims to change behaviours of health-/social-care professionals delivering the intervention by enhancing recognition of clinical uncertainty in clinical outcomes, patients in the last months of life and management of care expectations.

• Focuses on staff in primary, hospital and voluntary care, thus including different groups and organisational levels.

• Includes several complex intended outcomes, including changes in patient involvement in decision-making around situations of clinical uncertainty. The AMBER care bundle is tailored to the individual patient’s and their family’s need and circumstances.

Why research was needed

The AMBER care bundle was identified by NHS England as one of five key enablers of Transforming End of Life Care in Acute Hospitals,41,42 and it is currently being used across a network of approximately 40 hospitals including district general hospitals (DGHs). A further rollout is planned; however, recommendation 7 from the Liverpool Care Pathway for the Dying Patient has suggested that training programmes focused on clinical uncertainty and communication with patients and their relatives must be examined. 33 It was therefore imperative that a clinical trial of the AMBER care bundle took place to accurately quantify patient, clinician and health systems benefits, and that any harms are understood and managed. 43

Now more than ever, with ageing populations and increasing numbers of people dying from cancer and non-malignant conditions,44 health-care systems should provide every patient and their family a dignified death. 45 Facing deteriorating health and uncertain recovery is distressing for patients who may be dying and their families. This is particularly due to the frequent, possibly unnecessary, hospital admissions during the last year of life:46 in England, patients currently spend an average of 29.7 days of their final 12 months in hospital. 35 This is costly for health services and for society. These concerns are endorsed by the NHS Outcomes Framework,47 which includes two outcomes to improve care for people at the end of life: (1) the proportion of patients who die in their preferred place of death (PPD) and (2) bereaved relatives’ experiences of care. Interventional research, including an evaluation of the AMBER care bundle, aimed to understand how these outcomes can be addressed with this intervention.

Trial aim

The aim of this exploratory, multicentre, cluster randomised controlled trial (RCT) was to optimise the design of the AMBER care bundle, and to define the outcomes, for a fully powered definitive trial of the AMBER care bundle versus standard care.

Trial objectives

1. To examine recruitment, retention and follow-up rates at both patient and cluster levels.

2. To test trial data collection measures and determine their optimum timing in a larger trial.

3. To assess the degree of contamination at a ward level due to ‘between-ward’ staff and patient movements.

4. To provide a preliminary estimate of the clinical effectiveness of the AMBER care bundle compared with standard care to inform sample size calculation for the full trial.

5. To estimate the intracluster correlation coefficient and likely cluster size.

6. To examine differences in the use of financial resources between the AMBER care bundle and standard care.

7. To examine the extent to which the AMBER care bundle requires further refinement or adaptation (e.g. referral criteria to identify which patients would benefit most) to suit local conditions.

8. To assess the acceptability of the AMBER care bundle to patients, their families and HCPs.

9. To determine the ‘active ingredients’ of the AMBER care bundle that need to be maintained to ensure fidelity of the intervention for a full trial.

10. To assess compliance with and barriers to the delivery of the AMBER care bundle.

Objectives 1 and 6–10 pertain at both individual and cluster levels. Objectives 2 and 4 pertain at the individual level. Objectives 3 and 5 pertain at the cluster level. While assessing acceptability of AMBER, we explored potential benefits and intended harms.

Patient and relative interview demographics

In total, 24 interviews with 25 (including one dyad) patients and relatives were conducted (Table 11). Twelve interviews were conducted in each of the intervention and control arms of the trial. The reasons for non-participation are given in Appendix 7. In the control arm, six patients had a cancer diagnosis, compared with just one patient in the intervention arm. In the control group, seven patients and six relatives were interviewed. In the intervention arm, the majority of interviews were conducted with relatives. The relatively small number of patient interviews conducted in the intervention arm of the trial is related to a higher number of older patients, many of whom lacked the mental capacity to consent to be interviewed. The mean interview duration was 41 minutes (range 17–73 minutes). The interviews took place at various locations including the trial hospitals, a nursing home and the participants’ homes, and some were carried out over the telephone.

Non-participant observation characteristics

Non-participant observations of MDT meetings were conducted at three time points (at the beginning, in the middle and towards the end of the recruitment period) at each trial ward (Table 12). At three trial sites (Int1, Int2 and Con1), daily morning handover meetings were observed; at site Con2, discharge co-ordination meetings, which took place once a week in the afternoon, were observed. All meetings had a major focus on discharge. These meetings were attended by a range of HCPs, and their length varied. The majority of the meetings were led by a senior nursing staff member.

Focus groups

In total, four focus groups were conducted with HCPs; one was conducted per trial site. The characteristics of the HCPs who participated in the focus groups are presented in Table 13. Focus groups were well attended by a wide range of HCPs.

The inclusion criteria for the AMBER care bundle and implications for patient inclusion in the trial

Focus group participants shared their experiences of disagreement regarding patients they believed to be eligible for the trial. The reasons for this disagreement included confusion about the ‘middle ground’ between patients who were being ‘actively treated’ and those at ‘the end of life’. There was also a sentiment of a power disparity between doctors and nurses, and about how the final decisions were made for patients who could be approached for the trial. Specifically, at site Con1, the research nurse and ward sister explained experiencing difficulty in recruitment because the concept of clinical uncertainty was not fully understood by clinicians as being a legitimate entity in its own right. Often, doctors and nurses had contradictory views about which patients were suitable for the trial. The following two views exemplify this situation:

I think that’s been particularly difficult in this study. I think that is why we struggled to recruit. Because what we perceived to be a patient wasn’t certain (i.e. uncertain), was not necessarily the view of the medical team. I think it’s either ‘We are actively treating’ or ‘End of life’ [approving ‘hmms’ in the background from the other research nurses]. You know there’s no in-between.

Con1021, female, research nurse

Or the medics would say ‘No, they’re not’. But just listening to the handovers, it was like you’d identify everybody on the ward.

Con1018, female, ward sister

It was also evident from the research nurse perspective that despite attempting to explain their reasoning for patient inclusion numerous times, the medical staff were not engaging and had a different understanding of clinical uncertainty. A female research nurse stated the following:

It came to a point where we had explained it, and explained it. I think it got to a point where they said ‘No, no they’re not eligible’.

Con1021, female, research nurse

Other views, which were shared, drew on the more subtle and subjective characteristics of clinical uncertainty, which had implications not only for the inclusion of patients who might benefit from intervention but also for potential recruitment of those individuals to the trial. A consultant from site Int1 believed that the concept of clinical uncertainty should be open to more subjective assessment and was adamant that it was not an objective measurable concept:

It’s quite subjective, but that’s probably good. If you put strict criteria, you might miss some. Like we were saying, it’s almost like a feeling isn’t it, that someone is uncertain. It’s not this metric thing. You know, otherwise like this is a marker of uncertainty. I quite like the fact that the criteria is . . . uncertain.

Int1013, male, consultant

Issues around the inclusion criteria extended further to the risk of dying during their episode of care despite treatment. This criterion relied heavily on the ability to prognosticate, rather than clinical need, which was often challenging.

There was confusion about what represented an ‘episode of care’. In site Con2, a consultant shared his frustration in relation to his understanding of what an ‘episode of care’ implied, wanting to believe that the patients’ and families’ understanding of an episode of care extended far beyond the experience on a particular hospital ward:

My reflection is really, I think it’s just a bad expression; it’s the question of defining what the ‘episode of care’ is. It’s really what’s it came down to, wasn’t it? Initially, I was told that the episode of care finishes as the patient leaves the back door, which really isn’t true, is it? That’s the whole point of whoever is following their care to the community, as far as a patient is concerned. I’m hoping their perception of episode of care isn’t the ‘back door is closed, you’re in the ambulance going home and that’s it’.

Con2020, male, consultant

Concern about the ‘risk of dying’ criterion also opened up a discussion about variability in its perception across trial sites. The same consultant believed that different HCPs might have different interpretations about a patient’s clinical findings, which would have implications for estimating their risk of dying, rather than focusing on their clinically uncertain recovery. He implied that the trial should either focus on all patients where situations of clinical uncertainty exist or be more concrete about what risk of death entailed, and how this could be interpreted with high fidelity across different trial sites:

Yeah, he was very clear; he said probably 50/50 and that’s from his experience of because they’re using the AMBER care bundle in real life. So actually putting something like . . . well actually people will then think, actually there is a 50/50 chance of them surviving this episode of care, actually I would’ve found that much more helpful, because I look at their charts and I look through their organs, and the other things like scores and you can get mortality with the score and 20% chance of dying, is that fitting you or not really. How does that fit against the AMBER care bundle that is to be half and half, so it is difficult? My worry when we set out originally is what happens when that one of your study groups has an 80% death rate and the other has 20% death rate on the other site because they just slightly differently interpreted ‘the risk of death’. So I think being a bit more concrete about risk of death would be good.

Con2020, male, consultant

Modification of trial inclusion criteria

Patient recruitment in the control arm of the trial proved to be highly challenging. At site Con1, 55 patients were logged as ‘not eligible’ in the screening, principally because they were ‘not at risk of dying’ during their episode of care. Without appropriate education and training, research nurses and clinicians experienced enormous difficulty interpreting this criterion. Providing the level of education and support that was available to the intervention sites would have resulted in usual care being altered in the control sites. Consequently, there was considerable variability in the interpretation of this criterion, with the tendency for prognostication rather than consideration of risk. We therefore reviewed this inclusion criterion and how it was working at the control sites across a 4-month recruitment period with our PAG, TS DMEC and the principal investigators of both control sites. Because one of our feasibility objectives was to examine how the trial operated under field conditions and to make changes where necessary, we made a pragmatic decision to remove this criterion.

This change was adopted at the control sites after REC, HRA and local approvals for Substantial Amendment 6 were received (14 December 2017). We planned to monitor the effect of this change on recruitment. However, one of the control sites (site Con1) stopped recruitment owing to challenges around this inclusion criterion and did not have the capacity to reinitiate recruitment by the time REC, HRA and local research governance approvals were obtained. The other control site, site Con2 recruited eight more participants after the change in the inclusion criteria.

Trial settings

The control sites experienced a number of different challenges. It became evident from the principal investigator of site Con1 that consultant oversight of the trial ward was a key factor to take into account when setting up a trial. By the time the consultant on the ward had become familiar with the trial and its requirements, he or she was replaced by a new consultant who similarly needed to be introduced to, or reminded about, the trial:

Because there is a system where the consultant haematologist changes every week and there’s that five or six of them, aren’t there? So, they’re there every fifth week and you know, you happen to tell them every week about the study, remind them that the study is going on. Even the diabetologists change every, is it 2 weeks or month?

Con1023, male, consultant

Because identification of potential participants required clinical input, engagement of HCPs on the ward was crucial to recruitment success. A research nurse highlighted the lack of enthusiasm in relation to the trial from the HCPs:

For the patients, I don’t think there was negative impact . . . I don’t think the staff on the ward were keen.

Con1021, female, research nurse

Contradictorily, a research nurse located at one of the intervention sites witnessed enthusiasm for the intervention, and for her presence in relation to recruitment and data collection. This exceeded her expectations having come from a research background where the focus was not on older people:

So from my perspective, I’m mainly working in paediatric and reproductive health. So coming to care of the elderly, it was completely different and it gives me a unique insight into what’s going on the ward because I’m seeing it completely objectively. So, a few things to say . . . I was really, really impressed about how on board everybody was and everybody knew what AMBER was. On the whole, generally 90% of the time people were very supportive of our presence and what we needed to do.

Int1033, female, research nurse

Resource required for recruitment and consent

Focus group participants discussed the practicalities of identification and recruitment of potential trial participants and the emotive and complex nature of this trial. According to one research nurse at site Con1, they had hoped to access a larger team of research nurses to work on the trial. However, only a few nurses were willing to attend handover meetings and be involved in patient identification and recruitment. Other members of the research team were not comfortable working with a group of very unwell patients. For the research nurses who did collect data, it had been a tiring process:

To start off with as a research team we did it totally different. But then it was just two members of staff who came up every day, looking. A lot of the other staff felt uncomfortable working in this area is if we are honest. The two who did found all the end-of-life care tiring, even though it wasn’t a strictly end-of-life care study. The fact is that we got a paediatric nurse and a stroke nurse and they felt a bit out of their depth.

Con1020, female, research practitioner

Other research nurses drew attention to the challenges of recruiting patients who lacked mental capacity to consent. This was primarily because relatives who might have been able to talk about patients’ intentions regarding their participation were often not immediately available. The ramifications of this were that research nurses had to go back to the ward frequently:

The actual recruitment process was sometimes quite difficult because of what you said, the families not being here. Most of the patients we approached, we had to contact the relatives because the patients were too unwell. So being able to liaise with the families was difficult. We can’t be waiting for the families here, so we were called back about relatives that were here and popping back and asking the ward staff to tell us when they’re here. So the practicalities were difficult.

Int1033, female, research nurse

The non-participant observational work corroborated this concern, whereby a conversation initiated by an occupational therapist informed the rest of the ward staff that she had made numerous attempts to contact a patient’s wife but had yet to successfully speak to her.

Patients’ and relatives’ insight into uncertainty

On several occasions, research nurses spoke of the mismatch between some patients’ inability to comprehend their clinical situation even though they had previously talked in detail to clinicians about the context of the trial and their potential place within it. HCPs who had discussions with patients and relatives would assume that the individual understood that their recovery was uncertain. However, responses that research nurses got in relation to discussions with these patients did not always convince them that this was the case. The following female research nurse based at site Int2 stated the following:

I’m asking someone to do something because they’re on this AMBER care bundle, because they might get worse, or they might get better, and they say ‘Well no, I’m not going to get worse’. It’s not convincing for them to take part in the research project really. If they haven’t got full understanding.

Int2003, female, research nurse

A registrar located at site Con2 shared a similar sentiment, although this time it was focused more on the family members of patients who were very unwell. Although site Con2 had been reasonably successful in the identification and recruitment of patients, there were a few instances where it proved challenging to discuss a patient’s situation with family members as their willingness to accept the situation was critical:

Sometimes it’s like depends on their perception, sometimes you have a patient who’s far more advanced into the illness, whereby unable to give you a ‘Yes’ or ‘No’. So in those cases, you rely on family and it depends on their perception of what they think. So that’s where it . . . It sometimes becomes very tricky. But so far I think the ones who we’ve been trying to recruit, I think most of them have been quite successful. It’s probably one or two perhaps.

Con2019, male, registrar

Participant information sheet and consent forms

Comments here focused on the excessive length and detail of the participant information sheets and the consent forms. Although the development of the participant information sheets and other documentation was reviewed in detail by our PPI members, a research nurse stated that, from her experience, most individuals involved in PPI groups were notably different from the patient population we were aiming to recruit in this trial:

All the PPI groups I’ve ever dealt with are ex-professionals or current professionals, so they are very differently motivated group than actually in a bed, 80 years old, struggling to breathe, still got capacity, you know, you give that to an 80 year old.

You give a four-sided A4 to somebody like. It’s very different.

Int1003, female, research nurse

The nature of the trial documents and the manner in which the trial was sometimes explained to relatives were critical for its success. For instance, while the trial was being introduced to one relative, it became apparent that she interpreted this to mean that the patient was at the end of life. This caused her great distress. Although this was managed sensitively, and the individual was reassured that this was not the case, the principal investigator at the site exemplified the unexpected consequences as follows:

The daughter of the patient told me I was ‘Dr Death’ and ‘the Grim Reaper’! They were very upset about it and I think it was largely because they didn’t understand . . .

Con2020, male, consultant

The research nurse from site Int2 provided comments on the reality of conducting research with an inherently unwell and unstable patient group. Radical modifications to the level of detail of the trial documentation accompanied by a more streamlined process of explaining the trial and less burdensome consent processes should be considered. With these changes, the research nurse believed the trial had a much better chance of success, stating that many potential patients had been put off of the trial:

The consent process also needs to be changed. If you can get away some sort of ‘implied consent’ [would work better]. Because I’ll tell you what, this is a non-CTIMP [Clinical Trial of an Investigational Medicinal Product], so GCP [good clinical practice] is not legal, or the medicines so MHRA [Medicines and Healthcare products Regulatory Agency] doesn’t apply here. The law only applies to CTIMPs. So with this, there is nothing to say, you have to get a written consent form and I think you need to be pushing these boundaries with the ethics committees. This is why research in this specialty is not being done. You would’ve had dozens more questionnaires completed, dozens. Why can’t when I go in to a see a patient ask ‘Mr Smith would you mind answering some questions about your condition and how we’ve been treating you?’. ‘Yes, no problem.’ ‘OK.’ ‘Mr Smith you do understand that you don’t have to do this?’ ‘Yes.’ ‘You do understand that you can stop at any time.’ ‘Yes.’ So I can tick those and start asking questions. You would’ve had an 80% completion rate.

Int2003, female, research nurse

The design of the consent forms needs to be adjusted. There’s lots of improvement to make it easier to get people to consent. Not in a ‘We’re trying to fool people to get them into the study’. But you know reading the patient information leaflet for somebody who is very, very sick. It knackers them out and that’s just day 1. Day 1, talk to them. Day 2, reading the information leaflet. Day 3, I go back to do the consent form and they passed away or gone on to end of life, or they’re just too sick to even contemplate doing it. And it’s quite it’s a long process to trying to read somebody a 3–4 page long document, because they can’t read it themselves. They say read it to me, I get halfway through and they’re falling asleep because they are so, so sick.

Int2003, female, research nurse

The perceived impact of the trial among research nurses and patients

Ward staff who were responsible for potential participants’ care believed that they may have had some influence over patients agreeing to participate in the trial. Although some patients expressed interest in participating in the trial to HCPs who were caring for them, when research nurses followed up, patients declined to participate:

They didn’t want to upset me, because actually, potentially, I’ve done something to improve the quality of their life by giving them symptom control. So they’re thinking, ‘She’s done something for me, I’ll do something for her’. But when the researchers came along, whom they’d never met before, they said ‘no’.

Int2012, female, consultant

Although the research trial had been challenging for many of the research nurses, from consenting potential participants to data collection among an unwell patient population, one research nurse spoke emotively about the relationships she and her colleagues built with patients and/or their relatives during the trial. Unlike some of the studies that they had previously worked on, they felt that the focus of the trial provided them with a privileged position and opportunity to develop relationships with the participants. For example, the following research nurse stated that the perceived impartiality of the trial and her ongoing relationship with certain patients and family participants permitted them to talk frankly about their situation, which they did not feel able to share with the HCPs. When these nurses felt that it was appropriate, they actively encouraged these patients and families to communicate their concerns with the HCPs caring for them so that they could be served to better effect. A research nurse stated:

I think, the same as the other researchers, worked on the study, can’t be here today, he gave us a unique relationship with the relatives. So in a strange way, you’re in a unique position that they talk to you about things that sometimes they feel that they can’t take forward with certain ward staff. So we are able to encourage them into having those conversations making sure that those communications were taking place with the ward staff if they, the family has signed society, or certain query. So, whether because we were seen as external or whether we would able to form a relationship over questionnaires being done at different points, I just don’t know.

Int1033, female, research nurse

Patients’ views of being involved in the trial also highlight the impact of positive relationships being formed between research nurses and the participants as a reason for their continued involvement in the trial.

Pragmatic interpretation of the AMBER care bundle criteria

A critical issue associated with the delivery of the AMBER care bundle was the inclusion criteria. HCPs shared views on the inherent challenges associated with identifying patients using the criteria. At site Int1, two consultants discussed that their focus was on identifying ‘situations of clinical uncertainty’ in comparison with the criteria of being at ‘risk of dying during their episode of care despite treatment’. This suggested the need for simplified criteria:

There are some people that we’ve been trying to go down . . . sort of get them better but they’re not improving. I think those are the patients I kind of sort of see as uncertain or sort of unpredictable in this recovery. They’re the ones that we identify.

Int1013, male, consultant

I think most of our patients in the elderly ward are already coming in with uncertainty, because generally they’re aged, they’re frail, their history. So for me a lot of my patients would already be on AMBER before they’re even come in to hospital.

Int1018, female, consultant

At site Int2, the issue of who might benefit from being supported by the AMBER care bundle was constricted to the risk of dying. At this site, a female junior doctor explained that although many patients were unwell, with a possibility of dying within the next several months, unless a decision was made that this was likely to happen much sooner, they might not qualify to be supported by the AMBER care bundle. This reiterates the issue that the criteria do not focus exclusively on a patient’s clinical need but on prognostication:

Well sometimes it’s hard predicting whether they’ll die during this admission or when they’re going home. So for example, they might not die in this admission, but they are at the end of life in the next few months.

Int2019, female, senior house officer

Views shared by HCPs during the focus groups were corroborated by the non-participant observations of the MDT meetings that took place at sites Int1 and Int2. We observed a CNS asking her colleagues if they considered a patient as ‘being likely to die during their admission’ as part of their decision-making about whether or not the patient should be supported by the AMBER care bundle. During the same MDT meeting, two patients supported by the intervention were discussed in detail, yet the clinicians mentioned that they possibly had ‘a few more months left to live’. This was inconsistent with the AMBER care bundle inclusion criterion that states that patients should be at ‘risk of dying during their episode of care despite treatment’. We frequently observed that the clinical team made decisions about if a patient was going to be (or continued to be) supported by the AMBER care bundle not based on risk of dying but on their clinical morbidities, their disease progression (e.g. late stage of dementia/Parkinson’s disease), their likelihood of responding to medication or their ceiling of care escalation [e.g. if a patient was documented to be DNAR (do not attempt resuscitation), HCPs mentioned this as a reason].

Incidentally, the non-participant observations on the control wards identified that clinicians were practising ‘AMBER care bundle-like’ rules of care with some of their patients. For example, one doctor was heard to say, ‘If he [the patient] doesn’t respond to antibiotics, we should talk with his wife’. Others were observed as listing a patient’s morbidities and deciding at a MDT meeting that the patient was considered to be frail, with multimorbidities, and they should arrange to speak with the family. However, we did not observe instances in which patients discussed in this manner were also discussed as being at a risk of dying during their episode of care.

This concern about the AMBER care bundle inclusion criterion was also demonstrated in the account of a relative of a patient participant supported by the intervention. We identified a divergence between how this relative viewed their loved one’s situation in comparison to the HCPs who were caring from them, at least in relation to their decline. Despite the intervention’s intention to create awareness among HCPs around the risk of a patient dying during the episode of care, some relatives were surprised that HCPs did not see, or did not want to see, what relatives considered to be obvious. The daughter of a white British, 85-year-old patient participant with acute renal failure illustrated this tension:

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.

Int1017, carer

Reconsidering the daily reviews

The recommendation for the daily monitoring and review of patients supported by the AMBER care bundle was met with a critique. A consultant geriatrician at site Int1 clearly stated that he had accommodated the notion of uncertainty and wanted to instil it into his team. The uncertainty was very likely to be the normal state for many patients from one day to the next. Revisiting daily at the end of each shift to identify whether or not they were still suitable for the intervention, whether or not there had been any medical changes and whether or not a conversation had taken place with the patient and their family was therefore seen as unnecessary:

I think I prefer to have the team feel that there is always an element of uncertainty. It’s unlikely that tomorrow they won’t have any uncertainty. Because even if they recover, our patients they just go back quickly [agreeing chatter in the background]. So I prefer the review to be less often unless there is a drastic change in the person or you know sometimes people make a really good recovery.

Int1013, male, consultant

Building on the concern about the daily review, a junior doctor stated perceiving the overengineering of frequent reviews and sometimes frequent conversations to be distressing for patients and families. This may have been because some patients were frail, elderly and occasionally confused. It may also have been due to the nature of the discussions. Regardless, some individuals were interpreting the discussions differently, which might cause undue distress:

If you discussed it once and explained everything very well, the frequency of further discussions was possible I felt was too high trying to say that to them every day. Again the discussion can go from what we’re doing to make you better and then discussing the uncertainty again every day and it could be quite distressing because a lot of our patients are elderly and going through this with them, they just felt like ‘Oh that means I made progress. I’m well again.’.

Int2022, male, registrar

The concern surrounding the daily reviews also extended to the requirement of placing a sticker on the patient’s clinical notes in addition to recording their ongoing situation in the notes in detail. A male registrar said:

One thing I would dispose of are those stickers everyday about if it has been discussed with them, and there’s a prompt for discussing there.

Int2022, male, registrar

The HCPs’ concerns around the daily review sometimes contrasted with relatives’ expectations of how often they wanted to be kept updated about their loved one’s situation. Relatives suggested that when they came to visit their loved ones, they did not visit just to be with them but also because they wanted to get an up-to-date understanding of their loved one’s clinical situation. The daughter of a white British, 89-year-old patient participant with aspiration pneumonia illustrated this point:

Yeah, well I was there every day, so I spoke to someone every day. Yes, they kept me informed. When they were sending her home, they sat and did the form online with the nurse in charge of the ward. Every single day I was there.

Int1011, carer

For some relatives, visiting daily and seeking out someone to learn about any developments in the patient’s situation was far more challenging and led to frustration. The husband of a white British, 61-year-old patient participant with multiple sclerosis said:

No, it is a little bit in the dark. You got to go and find out yourself . . . The information is not volunteered readily. I think you got to go and find someone to try to explain it to you.

Int2008, carer

Concerns about the ‘labelling’ of the intervention

Some focus group participants raised concerns about the introduction of jargon, such as ‘AMBER’ to patients and families. At instances, family members were confused about the terminology and did not understand the use of the intervention in supporting their loved one. Being informed that their dependant was being supported by the AMBER care bundle became reminiscent of the now-withdrawn Liverpool Care Pathway for the Dying Patient,33 in which families became distressed. A junior doctor said:

We’d keep the family informed, but I think now it’s a bit confusing for the families like ‘Oh what is this AMBER bundle? Is my relative dying?’. And they just get overwhelmed.

Int2017, female, senior house officer

Competence and confidence in advanced communication skills

Focus group participants were aware that the support of patients whose situations were clinically uncertain often involved engaging in difficult conversations that demanded advanced communication skills. Although the nurse facilitator provided training on the rudiments of conducting difficult conversations, there were some HCPs, principally nurses, who were still hesitant about engaging in conversations. It was apparent that advanced training in communication was still required to deliver the AMBER care bundle. Consequently, some suggested that, in future, additional training should be made available to HCPs to enhance their skills and confidence in this area of care. A ward manager said:

So I would have that conversation anyway. I wouldn’t be frightened but I know but I know some of my nurses won’t at all. They wouldn’t talk with them . . .

Int2014, female, ward manager

Some of the HCPs’ hesitancies in having difficult conversations with patients and their families, and their need for more training in advanced communication skills, was also evidenced in the accounts of relatives of patients who had been supported by the AMBER care bundle. For example, during a period of a female patient’s acute deterioration, a doctor spoke with her husband about escalation decisions and tersely brokered information to him about her deteriorating situation. A nurse stepped in, realising how distressed he looked on receiving the news:

But I thought that could’ve been handled a lot better. Especially when he glanced over her and said ‘Well it doesn’t look good, does it?’. And I thought ‘What are you actually saying then? You’re saying that’s it?’ and he said ‘Anyway, I’ll give you a bit of time.’. And then a nurse stepped in and said ‘We’re not saying we’re giving up on her. We’re not at that stage yet. We are still gonna keep treating her.’. But the way it was described to me was that was it. They’ve done as much as they could do and if anything, else happens, that’s it. I didn’t think that’s handled well . . . Like I say, there was a lot of activity around her and I spoke, sort of thrown and get involved to see what was happening and then a doctor sort of said to me ‘We need to know your wishes on resuscitation. If she has a sudden collapse, you know stops breathing or health failure or something, do we resuscitate?’ and I thought well that’s a question. Right, straightaway, literally as soon as I walked in the door really and they said ‘You’re the next of kin. We can’t ask [Int2008]; she’s completely unconscious. What are your views on it?’. Well I said ‘I know what [Int2008’s] views would be’. But I thought that could’ve been handled a lot better. Especially when he glanced over her and said ‘Well it doesn’t look good, does it?’.

Int2008, carer (husband of a white British, 61-year-old patient participant with multiple sclerosis)

Prioritisations of patients’ preferences

The HCPs welcomed the departure from solely focusing on patients’ physical symptoms and placing more emphasis on engaging in early and important discussions with patients and their families about their preferences, and the inherent value of documenting these discussions.

Participants stated that they believed that conversations with patients whose situations were clinically uncertain were taking place earlier than they had previously taken place as a result of the intervention. Conversations now included a focus on patients’ and families’ PPC and PPD, whereas previously conversations primarily focused on physical symptoms. Moreover, they believed that the AMBER care bundle discussions not only respected patients’ autonomy but also placed them in a better position to realise their wishes. The following two HCPs, a physician associate and a consultant, shared these views:

It’s actually massively important to the patients, isn’t it? We didn’t appreciate it. Going more than patients want out be out of pain, they want dignity and I think at the end of life, they want control and some of that control is about the environment that they’re in. I think it’s about respecting that patient’s wishes and uhmm empowering them and give them their last wishes.

Int1013, male, consultant

I think it prompted conversations earlier. Because we recognise and we’re looking for these patients, we then initiate the conversations earlier than we may have done previously.

Int1002, female, physician associate

The relatives of patients supported by the AMBER care bundle provided accounts of how they perceived the care. There were instances where holistic care and open and compassionate communication were evident, which chimed with one of the intended benefits of the AMBER care bundle: a reduction in anxiety levels. This was typified by the following accounts from the relatives of three participants:

I think it was, erm, we were very impressed, the doctors worked well, they, yes, they spent a lot of time, in fact, in a way, erm, you could say too much time, because I think they were concerned that my father would be upset, so they had sort of prepared themselves for a very difficult meeting, interview, discussion . . . They were very caring in a sense, they were very concerned that he, you know, that, about how he might react, er, and, it went fine. It was good because the conversation was always involving him too, even though he was in bed or sometimes his hearing isn’t brilliant.

Int1001, carer (son of a white British, 98-year-old patient participant with prostate cancer)

They couldn’t have been more helpful. They really listened to us . . . When we had meeting with the doctors or whatever, they would say to us ‘do you have any questions?’ and they, everyone knows how short of time everyone in the NHS is but they appeared to have all the time in the world for us, which was very comforting. They were happy to, erm, answer as many questions as we had and to give us a full medical explanation . . . They were very open and helpful.

Int1005, carer (daughter of a white British, 84-year-old patient participant who had a heart attack)

Yes, I have to say that the doctors, one doctor in particular, she was very good, she would take us to another room and explain it to us and asked us if we had any questions. And she would stop to explain to dad and to us at the bedside. She was the best doctor there because she could communicate with you and she could come down to like dad’s level of comprehension.

Int1003, carer (daughter of a white British, 96-year-old patient participant who had a heart attack)

However, a number of relatives of participants were disappointed with the lack of appropriate and thoughtful discussions, which they felt were in their dependants’ best interests. The step-daughter of a white British, 92-year-old patient participant who had multiple fractures stated that her family’s views about the patient were never adequately canvassed. Moreover, when they were involved in discussions, the discussions appeared to be conducted in a manner that merely informed them about what was happening, often without consideration of their understandably heightened emotions. This was despite the family appreciating that what was planned was probably appropriate:

No, absolutely nothing . . . We weren’t asked anything about our preferences for her care in the future . . . We were told rather than asked by a consultant a week or ago that they wouldn’t uhm . . . they wouldn’t do anymore to her. Now, I felt I mean I agree with that, but I felt that he said it in such a way that ‘Well, we won’t do it’ [what we wanted]. Never mind what you’re thinking . . .

Int1016, carer

The daughter of a white British, 96-year-old patient participant who had a heart attack stated that although care was provided to her father, she felt that her concerns and questions regarding his care were often dismissed. Moreover, she believed that the HCPs avoided her and were not particularly interested in engaging with the family despite their daily presence on the ward:

You know they sort of ‘poo pooed’ what I was saying . . . The doctors took care of him and but there could’ve been more communication I think, really. We wanted more communication. We were there every day, so there was no reason why they did not stop and speak to us. They did at the end and realised that we were obviously interested in his care and what happened to him uhmm but yeah too many times, they could’ve stopped and spoken to us.

Int1003, carer

Involvement of health-care professionals in the decision-making process and teamwork

At the intervention sites, care and treatment decisions about patients with clinical uncertainty were made with the involvement of the MDT. A consultant stated that although it was the doctors who were perceived as being pivotal in the patient’s care, the views of other professional groups also contributed towards patient-centred decisions. This illustrates the acceptability of the ‘distribution of decision-making process’ among all HCPs, which is a component of the AMBER care bundle:

So it’s an MDT decision, because sometimes you know we as doctors, you know we go in, we see the patient on a ward but we’ve not been caring for the patient. HCAs [health-care assistants], everyone, everyone is involved in the decision.

Int1018, female, consultant

Instances of enthusiastic teamwork were observed during the handover meetings. At site Int2, we witnessed a range of HCPs contributing to discussions about patients’ suitability for the AMBER care bundle. This suggested acceptability of the identification part of the intervention and showed that the team understood its potential in helping patients and their families. This observation was slightly tempered by a number of HCPs, who required the nurse facilitator or the consultant to occasionally prompt them to consider potential patients who might benefit from the AMBER care bundle.

Joint acceptance and teamwork were also observed at site Int1. During a morning handover meeting, when the usual consultant was not available, a colleague from an adjoining ward stepped in to help. During this instance, we observed a number of HCPs informing him about the intervention and carefully explaining the various steps they took when making decisions about which patients were suitable for the intervention. Moreover, talking through the intervention components appeared to be natural and well rehearsed. During this discussion, a number of professionals volunteered information to the other team members about the progress of particular patients they were caring for. Moreover, they also told them they were trying to get hold of close relatives to explain to them the patient’s situation and condition.

Another example of enhanced teamwork was observed when the MDT made clinical decisions together and prioritised understanding the preferences of patients who were supported by the AMBER care bundle. A MDT (a physiotherapist, the ward sister and a consultant) discussed in detail the decision of whether or not to continue to actively treat an elderly patient. They also wanted to discuss the implications of this decision with the patient and learn more about his/her preferences. However, the consultant importantly mentioned that this could not happen before they comprehensively assessed the patient’s mental capacity (which is also highlighted in the AMBER care bundle tool).

Participants’ and their relatives’ perspectives on decision-making processes varied. We identified a number of instances where the views of focus group participants closely corroborated those provided by relatives of patients supported by the AMBER care bundle. For example, they spoke about the various ward staff being closely aligned in getting to know the patients, their thinking about a patient’s situation and their actions regarding care. They spoke of this as being akin to being ‘on the same page’. This is illustrated by the following two comments:

They all seemed to get on very well, yes. All on the same wavelength and saying the same things. Yeah when they sent her home there was an OT [occupational therapist], and he came to the house. The doctor phoned. The nurses made sure she had her cuddly toy with her. They all seemed to group together.

Int1011, carer (daughter of a white British, 89-year-old patient participant with aspiration pneumonia)

They did work well together, I felt, yeah it was a very nice atmosphere on the ward. I mean everybody, for example, the swallowing team and then physio[therapy], and then there was [name of the geriatrician] and [name of the physician associate], and mostly the same nurses who got to know mum, so, I think, there was definitely, erm, a good atmosphere.

Int1002, carer (daughter of a white British, 91-year-old patient participant with Alzheimer’s disease)

For a relative of a patient who had experienced less-satisfactory care in other parts of the hospital prior to their admission to the intervention ward, good teamwork was particularly welcome. The following account illustrates her sentiments:

Ward Y was particularly good from the point of view that they did seem to have some very good support of staff from the point of view of the physiotherapist, the occupational therapist there. It was quite a relief to go on to ward Y and suddenly everybody was planning what they were and how it was going, and um, you know a care plan was actually made for her. And so yeah from that point of view it was they kind of stepped up a notch.

Int1018, carer (daughter of a white British, 95-year-old patient participant with mixed dementia)

Nevertheless, there were times when teamwork was not evident and had negative repercussions for patient care. The daughter of a white British, 84-year-old patient who had a heart attack (Int1005C) mentioned that many of the HCPs had been very considerate and caring, exemplified by the comment ‘They couldn’t have been more helpful . . . very kind’, but she then explained that her father had been in receipt of unnecessary tests by the same HCPs who may not have been aware of the de-escalation for his clinical situation:

Sometimes we’d have people coming to do his blood pressure, erm, and his temperature and when I mentioned to this, to, erm, the consultant geriatrician at ward X, she was notably irritated because people had been instructed not to monitor him and they’d gone back to monitoring, I think, well for whatever reason, lack of communication.

Int1005, carer (daughter of a white British, 84-year-old patient participant who had a heart attack)

Relatives appreciated the challenge of clinical uncertainty but felt that more could be done to keep them in the loop of ‘not knowing’. In the following case, a relative had to take it upon themselves to seek out answers, often from different doctors. The husband of a white British, 61-year-old patient participant with multiple sclerosis stated:

No, it is a little bit in the dark. You got to go and find out yourself. I don’t . . . the information is not volunteered readily; I think you got to go and find someone to try to explain it to you. I suppose it’s a . . . just explanations could’ve been a bit better I suppose but . . . I think once they made up their mind that it was a chest infection, then everyone was heading towards the chest infection for treatment but before that it was a bit inconsistent because there was a lot of doctors with a lot of opinions. Yes, I can. I think that’s the bit that could have been handled better.

Int2008, carer (husband of a white British, 61-year-old patient participant with multiple sclerosis)

Inconsistencies in sharing information were at times frustrating for some relatives. Relatives appreciated that treatment plans may change as necessary; however, these changes in treatment plans were not always shared with them in a timely manner. In the following case, the wife of a white British, 70-year-old patient participant with corticobasal degeneration became confused by the mixed messages from various HCPs about her husband and began to question her own recollection of what had happened. This might have amplified her distress:

One would said one thing, and one would say another, and so it was a little bit like, what’s really happening? What’s going on? Erm, because if they had changed their minds and then I most probably hadn’t been told or heard, do you know what I mean?

Int2012, carer (wife of a white British, 70-year-old patient participant with corticobasal degeneration)

Although our findings from the focus groups and non-participant observations shed light on the acceptability of the AMBER care bundle, we did not observe the MDT meetings prior to implementation of the intervention. Therefore, it is difficult to identify the impact of the AMBER care bundle on teamwork and shared decision-making. Final decisions regarding patients’ care and their treatment are the consultants’ responsibility. Seeking information about patients and taking the MDT members’ views into consideration could depend on the individual consultant. The AMBER care bundle provides a platform for HCPs from various disciplines to share their input about patients whose situations are clinically uncertain. Our findings support the intended benefit of the intervention: enabling the sharing of information for decision-making among HCPs. However, we cannot confidently conclude whether or not the decisions were shared.

The value and simplicity in documenting discussions with patients

Participants expressed views on the value of talking openly with patients about their situation and having a clear plan. The devised plan is beneficial for other HCPs to understand and act on participants’ preferences for care, while acting as an aide-memoire for the person who devised the plan. In a busy ward environment, it could be easy to overlook or to forget important issues without a clear plan. The following core medical trainee voiced this view:

It makes you think about communication and especially when you’re busy and you focus on the medical aspect of it, you [then] remember that we promised the communication with the family.

Int2007, male, specialty registrar–core training 1

The documentation that accompanied the conversations was not perceived as a burden, which is important for HCPs who would be using it:

I think the documented discussion is very important. For the purposes of out of hours, we know the patient during the day and we have the conversations with the family and it’s there documented. At night, they know what the plan is . . . the escalation plan. I think that’s paramount really.

Int1018, female, consultant

This was amplified when HCPs could see that, overall, the document was simple to complete, a departure from many that had been used in the past. A junior doctor illustrates this view:

I think the simplicity of the paperwork itself is good. Because it actually doesn’t . . . we hate paper work. I think keep the simplicity of the paperwork.

Int2017, female, senior house officer

Delivery of care preferences

Where it was possible, the intervention required HCPs to engage in open conversations with patients about their situation and discuss their PPC and in some instances PPD. If HCPs were not honest during conversations about the feasibility of care and treatment options, patients may verbalise preferences that could not be delivered. This highlights the need for allocating adequate time for preparing for these conversations to ensure that the intervention is not causing distress. The following consultant suggested that conversations had the potential to disappoint patients when the system was not able to deliver:

We are . . . giving the patients the impression that we can give them what they want but then not being able to give them what they want. This may disappoint them.

Int1013, male, consultant

Acceptability of inclusion criteria

The HCPs questioned the acceptability of the current inclusion criteria. They felt that modification of the inclusion criteria from ‘patients being at risk of dying in the next 1 to 2 months’ to ‘risk of dying during their episode of care despite treatment’ meant that a reduced number of patients would benefit from the support of the intervention. Although the inclusion criteria are acceptable, alteration of the ‘risk of dying’ criterion might limit their utility. A female consultant stated:

Previously, when I used it before, this was what we talked about before, wasn’t it? [To EY] It was the 2 months or so and obviously they changed it, which meant that now we put less people on AMBER.

Int1018, female, consultant

Long-term acceptability of the AMBER care bundle

In order to operate and sustain the intervention over time, HCPs would need to make critical decisions about (1) who would be the ‘the AMBER care bundle champion’ and (2) what could be de-prioritised to create time for the bundle’s sustainability. HCPs considered the intervention acceptable while they received support from the nurse facilitator; however, they questioned the sustainability of the intervention without dedicated support. A female ward sister said:

The trouble is everyone has so many other responsibilities. Whereas the nurse facilitator’s role was to come in and to facilitate that. So she could focus on that. Whereas everyone else, if you’re asking someone else to take that role on, that’s not going to be their only focus. I feel like it needs someone to drive it.

Int1014, female, ward sister

The value in the identification of patients whose situations were clinically uncertain

The AMBER care bundle inclusion criteria were cited as being critical in prompting HCPs to actively consider patients on the ward who might benefit from the intervention. The following physician associate explained that, whereas previously this particular patient group had been absent from the clinical radar, now they had assumed a greater priority in the line-up of daily tasks:

I think for me it [the criteria] bumped it up the priority list, if that makes sense with jobs. Whereas before, it wasn’t highlighted as much . . . I think having conversations with families and updating them. I think what the AMBER care bundle really helped for me was the advance care planning side of doing things.

Int1002, female, physician associate

This finding was corroborated by the observations, where we witnessed discussion among MDTs about patients who fell into three main groupings: those who were acutely unwell, those who were deteriorating and, importantly, those whose recovery was uncertain. Regarding this last group, we often witnessed descriptions of patients as being ‘inbetween-y’ or those who were ‘up and down’.

The routinisation of active engagement and discussion with patients and relatives

Engagements with patients and families became routinised into the daily clinical practice of staff on the intervention wards. One female ward manager stated that although similar conversations with patients like this had previously taken place, the intervention prompted the HCPs to actively seek and prioritise these patients as part of standard practice:

We always recognised them but it became more embedded in our daily board round discussion about highlighting those patients and making a priority that we’re having those conversations but we would think to have those conversations with the relatives when they come in next whereas it’s be more of a we ask them to come in to have the discussions. So we were being more proactive about having those conversations rather than waiting.

Int1022, female, ward manager

We also witnessed instances when HCPs talked about patients who were supported by the AMBER care bundle, their PPCs and their families’ understandings of their situations. For example, we observed a conversation at site Int1 in which a HCP had discussed in detail the situation of a recently admitted patient with the patient’s daughter, who had unrealistic expectations about her mother’s recovery. Discussion moved on to how this conversation had progressed, taking on board the daughter’s views and what she considered to be her PPC for future care. Furthermore, at site Int2, when the nurse facilitator prompted the team to consider a patient to be supported by the AMBER care bundle, she discovered that the team had already identified the patient in question and had spoken in detail with her family.

Documentation of patient-centred plans

The documentation of conversations with patients and families and the development of a clinical plan were perceived to be an active ingredient of the intervention. This ensured transparency among team members. It was particularly useful for staff who worked out of hours, who would have otherwise been less knowledgeable about a patient’s situation. A female consultant and a male trainee doctor shared their perspectives:

I think the documented discussion is very important . . . Oh for the purposes of out of hours, we know the patient during the day and we have the conversations with the family and it’s there documented and this is what we’ll do if they become unwell or whatever. If they get called at night, then they know what the plan is, the escalation plan. I think that’s paramount really.

Int1018, female, consultant

The AMBER care bundle . . . sort of encourages you to have the discussions you wouldn’t have, and then it formalises it in a plan that can be followed by other colleagues. It’s there in writing and we’re requested to engage with that process.

Int2007, male, specialty registrar–core training 1

Documentation of plans and discussions is also intended to provide continuity of the information shared with patients and families by different HCPs. The following example highlights the importance of clear and detailed notes to ensure the correct flow of information to patients and relatives. The instance below highlights the importance of clear and detailed notes to ensure the correct flow of information to patients and relatives. It is here that some study participants suggested that the training that accompanied the AMBER care bundle placed enormous emphasis on understanding their concerns rather than being be delivered in a ‘tick-box’ manner. A white British, 65-year-old patient with motor neurone disease who was supported by the AMBER care bundle at the time of the interview stated the following when asked about the consistency of the information he receives:

OK, so it became clearer over time. So when you first got admitted, they weren’t that clear and then it got clearer?

Ask questions as we go.

So you ask questions as you go?

[He had a notepad that he used to communicate with the staff. He was not able to communicate verbally. He showed the interviewer all the questions he had asked before in his notepad.]

OK, so do you think that you’re getting consistent information from all the doctors and nurses or does it change or differ sometimes?

Occasionally different. Like a drug this morning. The nurse said no I am not for that drug. I questioned that. I was right.

Daily review and re-engagement with patients

Although initial discussion with patients and/or their relatives was considered to be an important ingredient in supporting their care and treatment, the subsequent re-engagement component of the intervention was also considered crucial. This took place daily and required the team to review whether or not a patient’s clinical uncertainty was still ongoing (whether they had deteriorated or improved). Re-engagement with the patients served two main purposes: checking the clinical uncertainty/AMBER care bundle status of the patient, which was fed back to the team for daily review, and following up about their requests, concerns and preferences highlighted during the initial discussion. Whereas follow-up conversations previously occurred on an ad hoc basis, the intervention required the HCPs to review and re-engage in conversations with the patient and their family throughout their admission. This helped the HCPs to understand how the patient’s situation and their preferences changed in the intervening time. The following core medical trainee stated that this drip-feed effect of continuing to address issues and engage with them had the benefit of improving overall communication:

So previously we might have discussed it with the family when they come in initially. Obviously, the admitting doctor is different to the ongoing team and then there may be ‘Oh someone else has talked about it, we won’t need to re-engage with it’, but I think with AMBER care bundle, because they have you have to keep discussing with patient, etc., it does encourage with this process of re-engagement with the family and therefore it probably does improve the communication with the family.

Int2007, male, specialty registrar–core training 1

The HCPs were also observed discussing the clinically uncertain recovery of the patients who were supported by the AMBER care bundle during their daily handover meetings. In the following instance, a patient who was currently supported by the AMBER care bundle had started to recover and the team were reconsidering mobilising him. The ward manager was noted to say ‘He’s no longer AMBER’ and then made sure that that the letter ‘A’ (indicative of being supported by the AMBER care bundle) was removed from the white board next to his name.

The role of the nurse facilitator

Although not considered to be a component of the intervention, a number of focus group participants spoke in positive terms about the critical role the nurse facilitator played in implementing the intervention both on wards and in supporting staff in its delivery to relevant patients and their families. The facilitator should be seen as an ‘active ingredient’ in the delivery of the intervention because achieving the intended benefits would be highly challenging, if not impossible, without her presence. According to the HCPs, she prompted them to think about issues that previously would not have been identified until the last days of life. She also supported the staff at critical moments, for instance when reflecting on the emotive challenges of having difficult conversations with patients and their families. The following male medical consultant and female ward manager explained:

I think the nurse facilitator showed that it works having someone like that. Two phrases that always stick to my mind were ‘preferred place of care’ and ‘preferred place of dying’ which is not something I would think about for a patient probably until we got to the discharge stage. Now, it’s sort of when I’ve seen the patient and identified that they’re AMBER, it’s almost uhm you know I kind of address that sooner so I guess that’s been a good thing.

Int1013, male, consultant

I was just going to say, having a dedicated facilitator helped the staff in having those conversations and supported them with the practical aspects of AMBER.

Int1022, female, ward manager

Our findings from the non-participant observations corroborate the views regarding the importance of the nurse facilitator in the delivery of the intervention. We observed at numerous points that the nurse facilitator encouraged and prompted HCPs to discuss patients’ day-to-day needs and to critically consider their clinical uncertainty. Typically, at handover meetings, the list of patients on the ward was briefly discussed by different HCPs. The following dialogue illustrates the role of the nurse facilitator in directing discussion towards thinking in what ways the AMBER care bundle might be appropriate for a patient. Patient 6 was briefly discussed and a junior doctor stated ‘We made him palliative’. This was followed by a discussion between the palliative care CNS and the junior doctor about whether or not they felt that he might die during that admission. After fielding the discussion between the two HCPs, the nurse facilitator summarised the situation by stating that the patient ‘sounded AMBER’ to her and an escalation plan was required. We also observed the nurse facilitator stating ‘She looks quite fail. She’s 90 years old. She looks AMBER, looking at her notes. COPD [chronic obstructive pulmonary disease] and vascular dementia.’. Other HCPs discussed in what ways the patient could be supported.

At other points during the same meeting, the fast pace of discussion was interrupted to focus in more detail on a patient’s situation. The nurse facilitator’s role in ensuring the intervention’s fidelity was evident: she was heard to say ‘Also still AMBER?’, to which the consultant replied ‘She’s not. She has definitely stabilised.’. A junior doctor stated ‘She came off [AMBER]’. The facilitator then reminded the team that they must now document the patient’s new clinical status regarding the AMBER care bundle in her notes. She was similarly noted to have told the team that when writing a discharge letter about a patient supported by the AMBER care bundle, they should remember to clearly state the patient’s clinical situation and the escalation decisions that had been made previously.

Overall, our findings suggest that a dedicated facilitator for the AMBER care bundle is needed to ensure that HCPs are trained, supported and prompted to consider all components of the intervention regularly.

Availability of relatives and health-care professionals

Participants in the focus groups stated that a large number of patients who had clinical uncertainty were elderly, confused and lacking mental capacity. Engaging in conversations about their situation with them was often challenging and not always possible. It was in the patients’ best interests that timely discussions took place with their relatives to better understand the patients’ preferences. However, relatives were not always available when required, and when they did come in, for example at the weekend, key staff familiar with their circumstances were often not present on the ward. The following ward manager made this clear:

Sometimes the families are not available. So a lot of the time we have some patients who come and their relatives don’t always come or they come in the weekend, or they come in the evening and then there aren’t the people [HCPs] there to have those conversations with them. So, I suppose the challenges on our ward, our patients have confusion or have dementia, and quite often those relatives are not on that same page necessarily . . .

Int1022, female, ward manager

The non-participant observations on the intervention wards corroborate this concern in relation to the unavailability of family members. For example, we witnessed a discussion about a patient’s wife who was receiving chemotherapy, and therefore would not be available to come in that day to have a discussion. In another case, a relative was not fully aware that her mother had dementia, and a significant time would need to be allocated to talk through the dementia diagnosis before it was reasonable to discuss her loved one’s situation of clinical uncertainty.

Communication skills and emotional labour

Focus group participants spoke in detail about communication being central in delivering important information to patients and their families. Communication skills are also critical for understanding patients’ concerns and worries and how they could be resolved. Participants believed that good communication was dependent on the possession of well-honed skills and confidence in having difficult conversations. Some consultants reported that they were experienced in this area, illustrated by the following comment:

It’s our bread and butter and we understand, we are better would you say, not better, better is not the right word but we’re a bit more . . . we can have those conversations.

Int1018, female, consultant

This was reiterated by other medical staff who felt that the intervention was acceptable as they already possessed the skills and knowledge required for the delivery of the AMBER care bundle components, such as the identification of clinical uncertainty and communication:

I think a lot of our . . . at least the medical side of it, a lot of our training has been around medical ethics, communication, breaking bad news, so it’s kind of drilled into us from the get go, I’ll probably say it’s a large part of our training.

Int1013, male, consultant

However, other HCPs, particularly nurses, stated that skills and confidence related to having difficult conversations were not necessarily equally shared. The following ward manager stated:

I think from a nursing point of view, for the junior staff to have some more kind of advanced communication training about you know breaking bad news and having those difficult conversations. Because for a lot of us it comes with experience and I don’t think you were saying in the doctors’ training there’s a lot of focus on it, the nursing training there isn’t.

Int1022, female, ward manager

The emotive context in which discussions regarding clinical uncertainty are required to take place was highlighted as a potential barrier to the delivery of the AMBER care bundle by some junior doctors. A ward manager reiterated this concern, suggesting that there is a need to engage in clinical supervision to support the staff involved in emotive discussions. This concern was challenged by the lack of protected time required to debrief and discuss what had potential to have a negative impact on the staff. The following two quotations illustrate these issues:

Broaching it with the family. It’s quite negative with the patients with uncertainty . . . we have to discuss the end-of-life care, DNACPR [do not attempt cardiopulmonary resuscitation]. So it’s hard to make it a positive thing that we are gonna keep with treatment and then nothing changes with taking care, if that make sense.

Int2023, male, foundation year 1

I mean it is a very emotive subject and you’re at the front of it. You know the relatives can take their frustration out on you uhmm so I think it is important that we all support each other with that and you know, having some kind of clinical supervision would be excellent. Because then as a team, we don’t ever have protected time to have a debrief, or discuss the difficult situations.

Int2022, female, ward manager

Some of the HCPs’ hesitancies around having difficult conversations with patients and their families, and their need for training in advanced communication skills, were also evidenced in the interviews with relatives of patients who had been supported by the AMBER care bundle. For example, during a period of acute deterioration, the husband of a patient recounted how tersely a doctor had spoken to him about his wife’s clinical situation and the consequent escalation decision.

Sample size

We aimed to recruit 80–90 participants throughout the trial. However, we were only able to recruit 65 participants. This limited our ability to identify differences in the candidate primary outcome measures, hence limiting the assessment of the preliminary effectiveness of the intervention. In addition, the small number of clusters included in the trial meant that we were not able to calculate the intracluster correlation coefficient required for a future trial. In future, we would extend the number of clusters included in the trial and implement learning from the screening and recruitment phase.

Inclusion criteria

Using inclusion criteria based on the eligibility criteria for the AMBER care bundle required subjective clinical judgement. This greatly limited the HCPs’ ability to reliably identify and then recruit patients. Consequently, we simplified the inclusion criteria at the control sites. Although we were not able to capture the effect of this change within the time frame of the feasibility trial, any future study examining this clinical issue would be required to adopt simplified inclusion criteria in order to improve the reliability of the identification of potential participants. Alternative recruitment strategies might also include recruiting and consenting all the patients within the clusters. Patients who fulfilled the simplified inclusion criteria in relation to their situation of clinical uncertainty could then be retrospectively identified from their clinical notes, an approach previously used in our comparative study of the AMBER care bundle. 48 However, the quality and level of detail of information contained within patients’ clinical notes can be highly variable.

Participant information sheets

The participant information sheets, although developed in close collaboration with our PPI members, require further improvement, particularly when attempting to engage frail older people. There were strong views from the research nurses that the documentation developed was overly lengthy, too detailed and too complex for our intended target patient population. Consequently, a number of potential participants were discouraged from engaging in the trial, preventing them from sharing their experiences. Little guidance exists in this research area in developing trial information documentation that is acceptable in both level of detail and length to both potential participants and research ethics committees. 99 Future studies working in this area of health care should be permitted to evaluate to what extent briefer participant information sheets convey adequate information about the trial and the potential benefits and risks to potential participants and the influence of this in relation to trial uptake. This could be compared with the current standard typically required by the HRA and research ethics committees.

Consent processes

Closely related to the above issue is the perceived burdensome manner in which advanced consent was obtained from patients, many of whom were frail, elderly and fatigued. According to the Department of Health and Social Care, informed consent can be written, oral or given ‘by implication’. 100 Therefore, it has been suggested that written consent is neither sufficient nor necessary for valid consent to be present. 101 Based on the previous section, we propose that a future trial incorporates a more pragmatic process of obtaining consent in which the patient participant explicitly agrees to what is proposed by the researcher, and the researcher then notes this. Again, this approach should be compared with approaches currently used to examine the relative acceptability of interventions.

Loss to follow-up

In our trial, the main reason for loss to follow-up was discharge from the trial ward. This limited our ability to utilise the data collection at the follow-up points. This finding was evident in both arms of the trial. As our data collection was limited to the time frame of patients located in the trial ward, we were not able to carry on collecting data from the participants. In response to this limitation, any future study examining this area of health care with a similar patient population should consider designing a study that either aims to recruit patients at an earlier point in their hospital admission or allows follow-up after discharge. A potentially more appropriate commencement point for recruitment within the hospitals could be acute medical units. Instances of clinical uncertainty and decisions regarding patients’ further treatment and care within the hospitals often take place in the acute medical units. This would extend the potential of collecting data at time point 2 for a larger number of participants who would otherwise have been discharged at this point.

Selection of trial wards

Guided by the AMBER care bundle development team, we made use of the ‘heat map’ approach to identify wards with the largest numbers of patient deaths. Consequently, we did not identify wards with similar specialties across the trial arms. This resulted in a case mix of patients who were quite different. The effect of this was most pronounced with the inclusion of care of the elderly wards that skewed the age balance between the trial arms. The mean age of the participants in the intervention arm was inevitably higher than that in the control arm. Future studies should not base the selection of trial wards solely on the number of deaths per ward, but also on other factors, importantly including ward specialty, the potential active engagement of ward staff and the presence of the principal investigator on the ward.

Selection bias

Although the opportunity to participate in the qualitative components was offered to all participants (patients or their relatives and HCPs), it is possible that those who chose to participate differed from those who did not. This may limit the transferability of qualitative findings to different contexts and has implications for conclusions made about the acceptability of the intervention. We aimed to purposively sample a range of patients and their relatives and HCPs. However, more time and resources should be allocated to approaching potential participants who might have opposing views about the trial and intervention to overcome these biases.