Recommended summary plan for emergency care and treatment: ReSPECT a mixed-methods study

Work package 1 (see Chapter 2)

The first work package is a qualitative study of ReSPECT decision-making processes, using observations of ReSPECT conversations and interviews with clinicians and patients/family members, and an analysis of the quality of ReSPECT form completion. These data are combined with interviews and focus groups with general practitioners (GPs) and other community staff regarding their experiences of the ReSPECT process (note that this was originally part of work package 4 but is reported with work package 1).

Work package 2 (see Chapter 3)

For work package 2, we carried out an interrupted time series (ITS) analysis using measures of process and survival outcomes for in-hospital cardiac arrests (IHCAs), using data from the National Cardiac Arrest Audit (NCAA), a survey and freedom of information request about use of DNACPR forms and other emergency planning approaches. The planned analysis covered 2 years before and 2 years after ReSPECT implementation.

Work package 3 (see Chapter 4)

Work package 3 was a retrospective observational study that comprised descriptive and regression analyses, using routinely collected data from adult acute patient medical records and the NHS Safety Thermometer audit.

Work package 4 (see Chapter 5)

The final work package (i.e. work package 4) provides a narrative summary of key findings from each previous work package (i.e. work packages 1–3).

Work package 1

• To describe the clinician decision-making processes behind ReSPECT form completion, including how, when and why judgements are made, their ethics basis and patients’/families’ understanding and experiences of the process.

• To explore the ethics basis and the experience of patients/families in the decision-making process.

• To explore GPs’ experiences of the ReSPECT process, including uptake and attitudes to the ReSPECT process in the community, and how the ReSPECT process transfers across the acute/primary care boundary.

Work package 2

• To quantify the effect of the introduction of the ReSPECT process on the frequency of, and outcomes from, in-hospital resuscitation attempts when compared with standalone DNACPR decisions.

Work package 3

• To present a descriptive summary of patient characteristics according to ReSPECT treatment recommendation and to conduct an analysis of whether or not a DNACPR decision, made in the context of an overall treatment plan, is independently associated with risk of patient harm.

Work package 4

• To synthesise the key findings from the study, to identify future research priorities from the patient, clinician and policy-maker perspective and to effectively disseminate findings, ensuring that key messages are integrated into future development work of the ReSPECT process.

A description of the context for implementation from regular meetings between sites, researchers and the ReSPECT National Working Group was planned as part of work package 4, on the recommendation of the Study Steering Committee. However, this work was discontinued because the ReSPECT National Working Group developed support and monitoring systems that would have duplicated this aspect of the study and this placed an undue burden on participants.

Characteristics of sites

Of the six acute NHS trusts participating in work package 1, three implemented the ReSPECT process between December 2016 and December 2018. We used purposive sampling for diversity according to volume of admissions, performance according to CQC banding, social class and ethnic mix of populations served, and approach to ReSPECT implementation and its uptake.

Four sites were teaching hospitals with general and specialist regional and national services. One site comprised three district general hospitals. All sites provided some specialist services to national and international patients, although these services differed in number and specialties. The numbers of inpatient beds in participating hospitals ranged from 450 to 1300. Inpatient activity ranged from over 100,000 a year to over 200,000 inpatient and day case admissions. Trusts with more than one acute hospital chose which trust(s) would participate in the study. During the study, two of our sites merged to form one trust (but retained their different approaches to the ReSPECT process during the study) and one trust merged with another trust not participating in this study. Our participating trusts had CQC ratings of either ‘needing improvement’ or ‘good’ from inspections around the beginning of the project in 2016–17. One trust had recently improved from a CQC rating of ‘inadequate’.

Sites covered urban (n = 3), both rural and urban (n = 2) and mostly rural (n = 1) populations. Populations served by participating trusts ranged from approximately 500,000 to > 1,000,000 people. Two trusts served areas with a larger, than average, ethnic minority population. Two trusts served areas with a larger, than average, white population. One trust served a mixture of areas, with some areas having larger, than average, ethnic minority population and other areas having a larger, than average, white population. Based on the Index of Multiple Deprivation (IMD) figures,36 two trusts served areas in which the majority of the population was more affluent than the average in England, whereas four trusts served areas in which the populations were more deprived than average. Childhood poverty rates ranged from almost 15% to 33% in areas served by the different trusts. Compared with the average life expectancy in England, life expectancy was worse for some trusts’ populations, but better in other trusts’ populations.

We were unable to select sites according to type of ReSPECT implementation, as planned, because of the small number of early adopting sites from which to recruit at the time. An overall summary of the implementation approaches used by sites follows.

Three sites changed from using a standalone paper-based DNACPR system to a paper-based ReSPECT system, and one of these sites also developed an electronic version during the study. One of the previous DNACPR systems was area wide (i.e. covering local acute trusts, community health-care organisations and the ambulance service). Another site used an area-wide DNACPR form and recorded escalation plans separately in the patient record. One site had its own electronic ECTP and changed to using an electronic ReSPECT form. One site has been using its own electronic ECTP for several years prior to implementing the ReSPECT process. Here, additional sections from the ReSPECT from were added to the site’s own ECTP on the electronic system and the site was piloting the actual ReSPECT form (paper version) on two wards. Adoption of the ReSPECT process in general practices in the areas served by our sites varied. In the two sites that had an area-wide DNACPR system and one other site, general practices across the area also adopted the ReSPECT process, but later than the acute trusts. At another site, the Clinical Commissioning Group (CCG) adopted the ReSPECT process a considerable time after the acute trust. At two sites, there was no area-wide plan to adopt the ReSPECT process at the time of the study.

All sites had a ReSPECT implementation lead. The professions of the ReSPECT implementation leads varied from resuscitation officers to medical or palliative care consultants. Working groups supported the lead at all sites to varying degrees, particularly when it came to running implementation activities. Some working groups included representation from community health-care providers and the ambulance service, and most working groups involved a range of clinicians from different acute specialties.

Five sites implemented the ReSPECT form on 1 day. At the sixth site, the adaptation to the site’s own ECTP was implemented on 1 day, and the pilot of the full ReSPECT form was conducted on two wards. The working groups carried out awareness-raising activities and conducted staff education for between 2 and 6 months before the implementation date. Most sites reported using e-mail, banners on trust intranets and attending meetings and giving presentations to raise awareness.

Staff education about the ReSPECT process included presentations, information tailored to different professions and, in some cases, advanced communication training for staff who would conduct ReSPECT conversations. One site emphasised the differences between its existing ECTP and ReSPECT as a focus for its training. Not many educational materials were available to the early adopting sites from the National Working Group and some sites reported developing their own. Some sites developed their own patient information leaflets. Some sites also used reminders and prompts to help embed the ReSPECT process. Audits were used to monitor progress and provide feedback.

Sites developed their own policies for the ReSPECT process, including policies regarding who could conduct ReSPECT conversations and who should take responsibility for what was recorded on the form. At most sites, a consultant was the senior responsible clinician, but at several sites a senior specialist trainee could be too. At some sites, junior doctors and senior nurses could initiate ReSPECT conversations; however, senior doctors had to review the process and sign forms as the responsible clinician. In other sites, junior doctors could sign ReSPECT forms as long as they were reviewed by a senior doctor within a specified time frame.

Interviewing clinicians, patients and families

We obtained written informed consent from clinician, patient and relative interview participants. Approaching patients and family members around the time a ReSPECT discussion occurred raised concerns about intrusion of privacy and causing further distress to patients and families who would already be anxious. The clinical team caring for the patient made the first approach and guided the researcher as to whether or not it was appropriate to approach the patient/family and, if it was appropriate, the timing of any approach. Interviews were tailored to individual patient and family member needs and could be brief. It was made clear that the interview could be stopped at any time if the participant wished to so.

To facilitate participation of people who did not speak English or who were deaf and able to use sign language, we made provision for a translator who was not a member of hospital staff.

Observation of clinicians engaged in making ReSPECT recommendations

We sought written informed consent from clinicians for observing their involvement in developing ReSPECT recommendations. The focus of the researcher’s observation was the clinician. However, the researcher was present when the clinician interacted with a patient or member of the patient’s family. At the start of any such interaction, the clinician introduced the researcher to the patient/family member and sought their permission for the researcher to remain, explaining that the researcher was there to observe the clinician. If the patient/family member did not want the researcher to be present, then the researcher would withdraw. The patient/family member could change their mind at any point without giving a reason and without prejudice. Verbal consent was sought from other staff present during observations.

For logistical reasons, it was not possible to obtain written consent from all patients and their families for the presence of a researcher conducting observations within a particular clinical area (e.g. hospital ward/emergency department). However, information about the study was displayed in these clinical areas and the researchers were clearly identified by their attire (i.e. a top with the word ‘researcher’ printed on the front and back).

Clinicians with whom the researcher had informal conversations regarding the ReSPECT process gave verbal consent. The researcher used pseudonyms to anonymise all participants in fieldnotes.

Accessing clinical records

The study required the research team to access relevant information from the patients’ clinical records and NHS Safety Thermometer audit data. Our approach sought to balance (1) respect for the patient’s right to information in their medical record being treated confidentially, (2) the risk to the validity and public interest in the research being harmed by a biased sample, and (3) consideration of practicable alternatives to obtaining consent.

This part of the data collection was the subject of our CAG approvals. CAG approvals allowed us to collect pseudoanonymised data on all eligible patients on participating wards from patient records without the patient’s consent. We provided study information leaflets that detailed how patients could inform study staff that they did not want their data collected or that they wanted their data removed after collection and before the end of the study data collection period. The rationale for CAG approvals is outlined in Report Supplementary Material 3.

Use of anonymised data from the National Cardiac Arrest Audit

Anonymous data from the NCAA were transferred in accordance with appropriate regulations and WCTU standard operation procedures on data security to the University of Warwick study team. The data were accessed by the study statisticians only and contained aggregated data with no individual patient-level data. The NCAA hold this information under CAG approval [reference ECC 2–06(n)/2009]. The NCAA Management Committee gave permission to access the data required for this study.

We planned repeated annual surveys to understand what systems were in place to record DNACPR decisions; however, after a poor response to the first survey, we gained additional approval to conduct freedom of information requests to NCAA participating sites. We needed a much higher response rate to conduct a robust analysis.

Ethics considerations

Patients and families provided verbal assent before each ReSPECT conversation was observed. Written informed consent was provided by clinicians who were formally observed and/or interviewed, by patients and families who were interviewed and by ward managers (in the last four sites) to state their agreement to contextual observations on their ward. Verbal consent was provided by clinicians who participated in informal conversations with the researcher during contextual observations. Families were interviewed if the patient lacked capacity, or with the patient’s consent if they did have capacity. All participant quotes presented in this report have been screened for identifying details. Participant identifications (IDs) indicate study site number and a random participant number, and cannot be traced to the participants.

Descriptive findings

Based on a thematic analysis of observed ReSPECT conversations, the following conversation typology was developed (adapted from Eli et al. 1):

1. Resuscitation and escalation (n = 31). Conversations in which the key aim is to record a recommendation about CPR and/or other elements of escalation of treatment [e.g. ventilation, intensive treatment unit (ITU) admission]. These conversations could be either open ended (i.e. exploratory; n = 13) or closed ended (i.e. persuasive; n = 18).

2. Confirmation of decision (n = 8). Conversations in which the key aim is to confirm with a patient and/or their family members whether they still agree with or would like to revise a previous ReSPECT or DNACPR recommendation.

3. Bad news (n = 4). Conversations focused on delivering devastating prognostic news. Although DNACPR and other escalation decisions are discussed, the prognosis is central.

4. Palliative/future care (n = 5). Conversations in which the focus is on decision-making regarding future hospital admissions, care in the community and transitioning to comfort care.

5. Clinical decision (n = 1). Conversations between colleagues, where only clinicians are involved.

Thematic findings

Working with the conversation types, we developed three themes corresponding to the research question ‘how, when and why are clinicians making ReSPECT decisions in the acute hospital setting?’:

1. why: planning for the possible and the inevitable

2. when: responding to hospital-based, clinical and patient/relative prompts

3. how: engaging with treatment options, patients and families.

The findings presented in this subsection have been adapted from Eli et al. 1 The adaptation retains the data and interpretations presented in the primary publication, but presents different illustrative participant quotes to avoid replication.

Clinicians are most likely to see community-issued ReSPECT forms come in with elderly patients

Clinicians were aware that ReSPECT conversations often take place when people move into care homes. Clinicians said that many of their patients with a community-issued ReSPECT form were admitted to hospital from care homes, and these patients tended to be ‘elderly with comorbidities’ (site 4, C07) or with ‘severe disability or advanced dementia’ (site 3, C05). Community-issued ReSPECT forms ‘almost always’ (site 4, C09) recordeDNACPR decisions for these patients.

Community-issued ReSPECT forms are generally useful, but recommendations may be questioned

When a patient arrives on a ward with a pre-existing ReSPECT form, the clinician first checks that the form is appropriate and valid, and this often involves a short conversation with the patient or their family to confirm that they are happy with the existing recommendations. Some clinicians reported that they have ‘to confirm that it still applies, every time the patient comes in’ (site 3, C07). Knowing the date that the ReSPECT form was issued was considered important, as a patient’s health can deteriorate rapidly:

. . . it’s good to know when they were written . . . the situation might be completely different now.

Site 4, C09

Forms issued in community settings were generally considered useful. Knowing that the patient had already had a discussion about possible treatments and expressed their values and preferences was said to be particularly useful for patients admitted into critical care, where intervention could carry significant burdens. Clinicians also reported that they find it useful to have information about ‘treatment escalation decisions beyond resuscitation’ (site 3, C04) and reasons for certain treatment decisions.

Two clinicians commented that community-issued ReSPECT forms are not useful. One clinician reported that only CPR decisions are useful and the other clinician said that they would ‘always have the discussion again’ (site 4, C04). Another clinician expressed concern that community-issued ReSPECT forms might include recommendations for treatment that they do not think clinically appropriate, which could create tension with patients’ expectations:

. . . as long as the form didn’t then ask me to give treatment that I didn’t think was going to be in their best interests, that’s my concern . . . I just would hate to have patient families’ expectations built even higher than they already are about having treatment that’s not likely to be in their best interests.

Site 3, C05

Clinicians said that they do not always adhere to the recommendations recorded on community-based ReSPECT forms. Clinicians said that they would question recommendations against certain treatments if they felt that these treatments would be beneficial:

. . . if it is actually going to benefit . . . and only will go on for a very short period of time then, yes, I would consider it.

Site 3, C06

ReSPECT decisions are not questioned if the patient was admitted to hospital because of the condition that the ReSPECT form addresses. DNACPR decisions are also not questioned:

. . . the resuscitation part, it is saying ‘do not resuscitate’, I would almost always appreciate that decision.

Site 3, C06

There are advantages to initiating the ReSPECT process in the community

Hospital doctors felt that GPs would be good at holding ReSPECT conversations, as GPs know their patients well and can ‘appreciate the [patient’s] medical comorbidity, the psychological support, the social infrastructure’ (site 6, C06). Some hospital doctors also thought that holding these conversations while patients were relatively well and comfortable [as opposed to in hospital when ‘they feel vulnerable’ (site 1, C03)] is better for patients. Other hospital doctors said that community-issued ReSPECT forms indicate ‘what is important for the . . . patient in terms of symptomatic versus curative intent’ (site 4, C02) and provided a prompt for discussions about ceilings of care:

. . . it won’t have detailed ceilings for all hospital treatments, but it will be enough to allow me to elaborate upon that . . .

Site 2, C04

However, one participant pointed out that, when people are healthy, they do not consider emergency situations:

. . . you don’t sit at home watching TV at night thinking, ‘You know, I only want to have resuscitation in such and such circumstances’. It’s not, it’s not what people think about.

Site 1, C02

There needs to be better sharing of ReSPECT recommendations across primary and secondary care

Participants agreed that all health-care professionals involved in treating a patient should have the same ReSPECT information. However, participants expressed confusion about whether or not community-issued forms apply when patients are admitted to hospital. The original format of the ReSPECT form (i.e. a hard copy held by patients) was criticised as being impracticable in emergencies, as ‘a lot of people don’t bring the ReSPECT form with them, they are in a rush to come to the hospital and they don’t remember’ (site 2, C01), and, likewise, for patients who frequently transition between primary and secondary care:

. . . we’ve got dialysis patients who’ve got ReSPECT forms in place. The recommendation is that they should carry that with them at all times. That’s not practical. What we need is something that is available to the community based health services and ambulance service, and the hospital.

Site 1, C07

Participants felt that a standardised electronic copy of the ReSPECT form that could be ‘attached to the patient’s electronic record and visible across all health care’ (site 1, C04) would more seamlessly transfer ReSPECT decisions.

The ReSPECT form is rarely used during conversations with patients and/or families

A paper ReSPECT form was used to engage with patients/families in only six conversations. In one conversation, the doctor showed a relative a blank ReSPECT form on a smartphone screen. In the two sites where digitised ReSPECT forms were used, patients/families had no interaction with the ReSPECT form during the conversation. Several clinicians at these two sites mentioned this as a downside of the digitisation of the ReSPECT process.

Most ReSPECT conversations elicit patients’ views on cardiopulmonary resuscitation and/or other treatment options

In most ReSPECT conversations the clinician asked the patient/relative about the patient’s wishes and preferences for CPR and/or other treatment options, including intensive care escalation and re-admission. However, in eight conversations, clinicians did not ask about patients’ preferences and, instead, informed them about the decisions the clinicians had already made about CPR and/or intensive care escalation. Clinicians took this approach when they were certain that these interventions would be futile, as one clinician explained:

. . . it was pointless to have a discussion knowing that you would not be able to offer the patient the options they might want.

Fieldnotes, site 1, C09

When clinicians asked about the patient’s wishes and preferences, the approach taken was either persuasive or exploratory. In persuasive conversations, although clinicians asked patients or their families what they would want, clinicians would also voice their own recommendation, explaining that CPR and/or intensive care escalation would be futile, sometimes using emotionally evocative language (e.g. telling patients that CPR would be undignified or would distress their families). In exploratory conversations, clinicians explored patients’ and families’ wishes openly, allowing these wishes to guide the rest of the conversation and the development of treatment recommendations. For example, after breaking bad news to a patient’s partner, an intensive care consultant said ‘I’m sorry to have to warn you about this but in this circumstance we do ask you as the person closest to her if you would know what her wishes would be in this situation’ (fieldnotes, site 4, C02). The patient’s partner asked for resuscitation, but not at the expense of patient’s quality of life, and the consultant responded with ‘we will resuscitate her’, but ‘there is a possibility it would not be successful . . . after repeated attempts we will step back and allow things to take their course’ (fieldnotes, site 4, C02).

In several cases, exploratory conversations resulted in no recommendations, as patients found it difficult to engage in the conversation, expressed ambivalence or asked for more time. In these cases, clinicians explained that they would either revisit the conversation later or leave the conversation where it was, hoping that it would encourage the patient to consider treatment escalation and have future conversations about this in community settings or with their families.

During ReSPECT conversations, many clinicians explain what treatments would and would not be provided

Clinicians spoke about what treatments (other than CPR) would and would not be provided in more than half of the observed ReSPECT conversations. The level of detail, however, was variable. In conversations conducted during ward rounds, treatments were typically mentioned in the context of the patient’s general management, rather than directly linked to what clinicians conceptualised as the ReSPECT conversation, which focused on CPR and/or intensive care. In conversations conducted outside ward round contexts, treatments were typically mentioned in the context of the ReSPECT process. In these cases, clinicians employed a reassuring tone, explaining to patients that they will still be treated regardless of a DNACPR decision, and this could be framed broadly [e.g. ‘it’s doesn’t mean we’ll stop treating you’ (fieldnotes, site 4, C09)] or specifically {‘[the doctor] said they will give her i.v. [intravenous] antibiotics and fluids, and fluid resuscitation, but ‘won’t bring you down to ICU [intensive care unit]’ (fieldnotes, site 4, C08)}.

However, in many (n = 21) of the observed ReSPECT conversations, clinicians did not speak about treatments other than CPR and/or intensive care escalation.

Patients and families are involved in ReSPECT conversations, but not always in decision-making

Of the 49 ReSPECT conversations observed, 48 conversations involved patients and/or their families and one conversation was a discussion between clinical colleagues for a patient who lacked capacity, had no next of kin and no friends, and for whom an independent mental capacity advocate could not be located. Conversations varied in the extent to which patients/families were involved in the decision-making process. Twenty conversations were open ended, with clinicians actively eliciting the patients’ wishes and preferences, and taking these into account in decision-making. However, in 20 other conversations, clinicians took a persuasive stance, using conversational scripts to sway the conversation towards a decision aligned with medical opinion or simply informing the patient/relative about a medical decision that had already been made. For example, a doctor was observed saying ‘should the situation worsen, we will not put [the patient] on a machine’ (site 4, C07) to a relative. The remaining conversations were confirmatory, with clinicians asking patients/families if they agreed with a previously issued ReSPECT decision or community DNACPR.

Clarity and uncertainty about patients’ trajectories affect whether or not patients/families are informed or involved in decision-making

As reported in our earlier publication,2 using clinician interview data, a clinician’s clarity or uncertainty about a patient’s trajectory often determined the extent to which the patient/family was included in ReSPECT conversations. When clinicians were uncertain about a patient’s trajectory, the ReSPECT conversation emphasised patient preferences to a greater extent:

. . . the greyer . . . the decision then the more and more you allow the patient’s own perspective to sort of sway the ultimate outcome.

Site 1, C09

Similarly, although clinicians were certain that palliative patients were nearing the end of their lives, ReSPECT conversations acknowledged clinicians’ uncertainty about how patients wished to spend their last weeks or months, thereby emphasising patient preferences in certain aspects of care:

. . . my only main aim was, to get out what she wants from the rest of her life, and how she wants to be treated and how she wants to be, like, managed in all her, including her end-of-life care and everything, so . . .

Site 6, C07

By contrast, when a patient’s illness and treatment trajectory seemed clear, then clinicians tended to take a persuasive approach to ReSPECT conversations:

What I don’t think’s particularly helpful is having a discussion with a patient where it’s very clear from a medical objective point of view that it would be futile to escalate treatment to give them the impression that you’re giving an option of escalating treatment. Because again, I don’t think that’s, that’s fair. So if it’s very clear then I think our job is to explain why we’re making these decisions carefully and sensitively.

Site 3, C05

Accordingly, some clinicians said that ReSPECT conversations were mainly focused on informing patients/families about what was possible medically, rather than seeking their involvement in decision-making. Several clinicians stated that it was important to clarify to patients and, in particular, families that clinicians were not asking them to make medical decisions:

. . . you don’t want the family to feel that they have decided . . . that’s the last thing you want, the last thing you want is for them to feel responsible for having made that decision.

Site 4, C10

Some clinicians advocated a middle ground approach. Although clinicians did not ask patients/families to participate in medical decision-making, clinicians used ReSPECT conversations to contextualise the eventual clinical decision:

It’s a shared decision . . . we give the medical information. We take on board the family’s values and preferences, and we issue a recommendation based on that. And I think that’s important rather than feeling, letting the family feel either that they’ve got to make the decision, and we’re washing our hands of it, or that we’re making the decision for them doesn’t matter what they think.

Site 5, C08

The ReSPECT decision-making process is shaped by the patient’s autonomy

Many of our patient and family interviewees expressed their preferences after their doctor had explained CPR to them, and believed that their doctor respected their wishes. One patient reported disagreeing with their doctor who suggested a particular treatment, and some family members felt that the doctor would have implemented their wishes, even if they disagreed with the clinical recommendations:

I think she would have listened to us, and done everything she could have to have implemented any other action that we insisted upon.

Site 6, F02

In contrast, other patients and families felt that their doctor simply informed them about the ReSPECT decision without discussion or explanation, as one patient said:

The doctor told me . . . ‘I hope you know we won’t, won’t resuscitate you’.

Site 2, P03

This patient went on to say that they agreed with the doctor’s decision, but also felt unable to disagree:

I’m quite prepared to go along with that. Well, I can’t do anything else, can I?

Site 2, P03

Several patients commented that they trusted the doctor to make the right decision on their behalf, as doctors are experienced, highly trained and will do whatever is in the patient’s best interests:

. . . to be honest, I always say, ‘I’ll leave it in your hands’. You know, they know best.

Site 1, P05

Many ReSPECT discussions were held while the patient’s family were present. Patients said that their family provided emotional support, but that the decision was theirs alone to make:

. . . my daughter . . . she says, ‘It’s your decision, mum. I’m here to support you, whatever you decide’.

Site 2, P06

One patient, who felt shaken by the ReSPECT discussion, said they did not want to make any decisions without their children, but was worried about upsetting them.

If a patient lacked capacity to make their own decisions, then their family was responsible for expressing their wishes. These family members said that the ReSPECT process should reflect the patient’s wishes:

. . . you do have to stop and think actually what that individual wants, not actually what you want.

Site 2, F02

A few patients had discussed their preferences with their family prior to their hospital admission, and this made the ReSPECT conversation easier for their families:

. . . my dad’s always, sort of, you know, very open about what happens, you know, when his time comes, . . . I know exactly what they want the end of their life, so, yeah, so, you know, we have talked about things like that.

Site 2, P02

Patients and families do not always recall or understand their ReSPECT conversation

Although some patients had good recall of the conversation, several patients could not clearly remember what was discussed. For example, one patient said there had been no discussion of resuscitation:

It hasn’t been spoken about. That word ‘resuscitation’ has never come into it.

Site 1, P01

Lack of recall could be due to the confusion of being in hospital, or being very ill or in pain at the time of the conversation [e.g. ‘I was in a bit of pain anyway, you know’ (site 1, P04)]. Family members too described having the conversation while feeling anxious and upset [e.g. ‘we were all a bit uptight, you know, because we’d . . . had the ambulance journey’ (site 6, F01)] or in the time of crisis:

. . . when you come in [to hospital] so much is going on, it’s something’s happened all of a sudden . . . you’re trying to focus on your [relative] . . . you’re trying to work out what’s going to happen to them.

Site 2, F02

Having ReSPECT conversations with ill patients or in an emergency was criticised by one person, who thought that this timing was inappropriate:

. . . when you’re really poorly, you know, it’s not a good time. ’Cause then you feel like giving up. So you need to be well for such conversations.

Site 2, P01

Some patients, and most families, felt they had understood the ReSPECT conversation, and that the purpose of ReSPECT had been clearly explained to them. There were some misunderstandings, however. One patient initially thought that the conversation was related to organ donation, then assumed that much of the conversation was to prevent litigation if the patient received treatment they did not want, and this patient’s doctor ‘made it clear that it wasn’t’ (site 1, P04).

Patients and families are unprepared for the conversation

Only one patient had expected to have a ReSPECT discussion during their hospital stay, after experiencing multiple previous conversations with doctors about resuscitation. Generally, however, patients and families were unprepared for this discussion. The unexpectedness of the conversation confused some people:

. . . my mind was kind of catching up with where he was going,’cause I didn’t, there was no real lead into it.

Site 6, P01

Some patients felt unable to think through their wishes or questions:

And now, looking back, now I’ve had time to think, there’s lots of things I would have asked, which I didn’t.

Site 2, P01

One person suggested that the ReSPECT process should be a two-stage process, with an initial conversation/leaflet signposting what the patient would need to think about:

. . . to have reasonable input from someone like me, I need to know we’re going to have that conversation, maybe even with a bullet list.

Site 6, P01

Another patient said that it was important to have time to discuss the philosophical aspects of life and death decisions with someone of faith.

People who were unprepared often felt upset after the conversation:

It shocked me a bit, actually. Very shocked and a bit frightened . . . I felt quite brave at the time and thought that I could cope with it, but as soon as the doctor went I wanted to cry.

Site 2, P01

The conversation left some patients/families concerned and confused as to their future trajectory. For example, one patient was told that they were having the conversation because of repeated hospital admissions that left them concerned that they were dying. Another patient reported that the conversation made them realise how quickly their health had deteriorated:

I never thought it would go downhill this fast.

Site 1, P03

Several patients and families said that, although upsetting, it was important to contemplate treatment planning, and one patient said that it was ‘marvellous’ (site 1, P03) tthat hey were able to express their preferences:

I know it’s an upsetting conversation and I did get upset, but it’s a conversation you need to have.

Site 3, F06

Participants mentioned that the transferability of the ReSPECT form was crucial so that everyone involved in a patient’s care (GPs, hospital staff, etc.) knew the patient’s wishes.

ReSPECT discussions focus on resuscitation

Many participants said that resuscitation was the focus of ReSPECT conversations. Patients and families recalled how doctors described the consequences or side effects of CPR:

I knew that if they had have done CPR, like, it probably would’ve broken her ribs and done a lot of damage, . . . And I just didn’t want it for my mum.

Site 2, F01

Although nearly all participants recalled discussing CPR, very few participants could recall the doctor talking them through other available treatments:

. . . no future treatment was talked about.

Site 1, P02

. . . we didn’t mention anything about intensive care or anything like that because I don’t think he had the time.

Site 2, F02

During the ReSPECT discussion, some patients sought reassurance of a peaceful, painless death:

You just hope it’s going to be peaceful but I was told that you wouldn’t suffer if they, you know, they wouldn’t resuscitate but, but I, I wouldn’t suffer. Well, which is what everybody wants to know, you know?

Site 2, P04

Concern for family shaped some patients’ views on resuscitation. One patient declined CPR because they did not want to place the responsibility for making resuscitation decisions on their son’s shoulders:

I didn’t want to put my closest people, you know, through any trouble . . . I didn’t want my son to have to go through it all, you know, just make a decision like that.

Site 2, P07

Several patients reported that they did not want to be resuscitated because they were elderly, saying that they could die at any time, did not want to burden their family and feared being left in worse health.

Family members described deciding not to attempt resuscitation as being in the patients’ best interests. The lack of quality of life a patient would experience after resuscitation influenced the decision to decline it:

I just think she would’ve just deteriorated and deteriorated and deteriorated, and I don’t think she would’ve ever come out of the bed, I think, after having that done, to be honest.

Site 2, F02

Families described how previous resuscitation decisions helped make the ReSPECT conversation easier, and these participants said that if they were aware that the patient would want to decline resuscitation then they felt obliged to ensure that decision was made, regardless of how hard it was for them:

I felt terrible saying it because it’s my mum and, you know, I don’t want her to go . . . I know that if she does go then she wouldn’t want to be brought back, and she’s already sort of, like, stated that she didn’t want to be resuscitated if anything happened. So I had to ask the question.

Site 2, F01

Families drew on previous experience when patients had repeated admissions. Conversations held on first admissions tended to be longer and more in-depth than follow-up conversations:

. . . we’d already made the decision ’cause I said to the [doctor] that [she had] already been in before, we’ve been through it again. So perhaps he didn’t feel like he needed to go through all that detail.

Site 2, F02

Doctors’ communication skills are both praised and criticised

Doctors were often praised by patients and families for their good communication skills. Participants felt able to ask the doctor questions and express their preferences:

. . . he listened, he didn’t sort of, like, talk over, he let us, sort of, like, say what we wanted to say . . . he made it, sort of, like, quite easy.

Site 2, F01

Doctors clarified what the current state of the patient’s health was, what might happen, and what treatment they would or would not receive if they deteriorated. In communicating about difficult medical decisions, doctors tended to speak clearly and honestly. Although participants appreciated this honesty, the honesty shocked participants who did not expect a resuscitation discussion:

. . . she told us straight which was her job to do. It’s no good beating about the bush if you’re doing, if you’re talking about life or death . . . it hits you hard, straight away when you get it straight in your face like that.

Site 6, F01

Other participants recalled that the doctor warned them that the conversation was going to be difficult, and they found this warning useful in preparing themselves.

Patients and families often praised doctors for being calm, kind, caring and compassionate:

. . . he was very sympathetic and very understanding, and he, sort of, put me at ease and everything.

Site 2, F01

Doctors also reassured families that they had made the right decision for the patient:

Doctor went over and said, ‘I think you’re making the right decision’, you know, just to give us, just to give me peace of mind.

Site 2, F02

Alongside this praise, participants acknowledged that these conversations were not easy for doctors:

I think it’s difficult for doctors. . . It is a horrible thing to have to tell someone.

Site 1, P03

However, not all participants were positive about their doctors’ communication skills. One person commented that doctors do not speak in patient-friendly language:

. . . doctors will talk doctor-talk.

Site 1, P01

Another patient felt that their doctor was too blunt in how they introduced the topic and ‘they could have been more subtle, couldn’t they, I think, not just come straight out with it’ (site 2, P03).

Patients and families commented that ReSPECT conversations could be improved by clinicians allowing more time. Several participants commented that their conversation was ‘short’ and their doctor was ‘a bit quick’ (site 1, P04). The lack of time meant that doctors did not discuss the different issues in detail. Some family members reported that they had to wait at the patient’s bedside until the doctors had time for a conversation.

Patients are unaware if or how the ReSPECT process affects their subsequent care

Patients did not discuss if or how the ReSPECT process affected the care they received. One patient was concerned that if they engaged with the ReSPECT process then they would receive different treatment, or no treatment:

I was just a little bit worried that if I sign up to this ReSPECT that I might not get the treatment that I need to keep me alive. Will it just be assumed that I’m going to die and just leave me. So that’s a little bit concerning . . . I’m getting the impression that there’s going to be different treatments for different people.

Site 2, P01

The patient’s condition is central to initiating the ReSPECT decision-making process

Most doctors said that they initiated ReSPECT conversations with patients who were at risk of deteriorating during their current hospital admission, and these conversations usually focused on making decisions about resuscitation:

I think when someone’s acutely unwell their main priority is, is discussing resuscitation and ceilings of, of treatment while they’re unwell and once they’re stabilised, that gives you time to have these further discussion to explore their wishes for their future care.

Site 3, C04

Accordingly, doctors often deferred ReSPECT conversations with patients who were acutely unwell, but deemed to be stable:

I didn’t think that she was someone who was likely to have a cardiac arrest that morning, the following day.

Site 2, C05

By contrast, an intensivist said that with ‘critically unwell’ patients the ReSPECT conversations were ‘repeated often on a daily basis . . . And if there’s a change . . . that particular ReSPECT form, the printed version is crossed out and, and is updated’ (site 4, C02).

Although ReSPECT conversations with patients who were deteriorating were often persuasive, uncertainty about a patient’s prognosis often led to ReSPECT conversations with a shared decision-making stance. For example, a consultant explained why a particular patient was asked about his intensive care preferences:

And he is one of those borderline patients that will sometimes be accepted by ITU and sometimes not. And my decision about whether to escalate him to ITU would be based very much on his preferences, so I had to ask.

Site 2, C04

Clinicians also initiated ReSPECT conversations with patients who had terminal illness, where the focus was not on acute deterioration, but rather on ascertaining the patients’ wishes for their end-of-life care:

. . . he’s got a terminal diagnosis, we’ve exhausted all the treatment options and all our focus now is on, is gonna be on his, on his symptom control. And so, the process of ReSPECT is, is the all-important question: ‘What’s important to that patient right now?’ . . . to make future care plans about whether a return to hospital is the right thing, what would be his preferred place of death . . .

Site 3, C07

The foci of ReSPECT conversations and decision-making differed according to the type of ward where conversations took place. In geriatric wards for longer-term patients, ReSPECT conversations tended to focus on broader aspects of treatment planning, involving patients and families in decision-making about palliation and re-admission. However, in acute geriatric wards (e.g. frailty-focused medical wards), ReSPECT conversations tended to focus on decision-making concerning CPR and/or intensive care. As an emergency medicine consultant explained, although the ReSPECT process was a shared decision-making process, during an acute admission, there was limited scope to explore patients’ wishes beyond resuscitation:

. . . is an acute situation the right time to have that [conversation] when they’re, they’ve got what’s almost certainly compromised mental capacity?

Site 5, C08

Focusing on limited conversations with acutely ill patients, as noted in our data, responds to the ethics duty to protect the patient from harm, for example from unsuccessful CPR or invasive interventions in intensive care unit (ICU), although possibly at the expense of articulating and recording recommended treatments that will benefit the patient. In the context of less acute patient conditions, ReSPECT conversations were more likely to be shaped by considerations of patient autonomy and wider consideration of benefits, with clinicians expressing a concern to elicit the views and wishes of the patient, particularly in situations of clinical uncertainty.

Clinicians vary in their conceptualisation of ReSPECT as a shared decision-making process

Differences between doctors’ approaches to shared decision-making reflected different ethics stances. Many doctors spoke of ReSPECT decisions as ultimately medical, saying that, although patients and families should be involved, the final decision is based on clinical judgement. A key reason for taking a persuasive or prescriptive approach to ReSPECT decision-making was to avoid patient harm. When doctors identified an intervention as futile, they limited discussion about it. For example, a respiratory consultant who supported shared decision-making in some ReSPECT conversations drew the line where futility was involved:

What I don’t think’s particularly helpful is having a discussion with a patient where it’s very clear from a medical objective point of view that it would be futile to escalate treatment to give them the impression that you’re giving an option of escalating treatment. Because again, I don’t think that’s, that’s fair.

Site 3, C05

To avoid situations where patients requested treatments that doctors deemed futile, doctors employed conversational strategies that limited patient involvement. A junior doctor working with a largely elderly patient population used a conversational script that positioned CPR as ineffective in elderly people before asking patients about their wishes. Without this script, the doctor explained, awkward situations arose:

But I have had conversations where I would say, ‘Have you had any thoughts about CPR?’. Sometimes. if I leave it open, I find their response would be, ‘Oh yeah, I’d, I’d like it, please’. And then he becomes . . . obviously it’s a little bit more difficult to walk it back and say that, ‘No, I don’t feel that that’s appropriate’.

Site 4, C08

Notions of futility were closely linked to weighing the harms and benefits of particular interventions. When harms were deemed to outweigh benefits, doctors informed patients and families about their decision, as opposed to asking patients and families to participate in decision-making. A consultant described using ReSPECT discussions to explain medical reasoning about harms and benefits:

. . . it’s when the risks and benefits and burdens outweigh the benefit then you shouldn’t give that treatment. And it has to be black and white like that for them to understand the decision-making that we’re trying to logically go through.

Site 5, C03

The balancing of harms and benefits was not limited to individual interventions and individual patients. Some doctors were concerned about the harms that a futile intervention might pose to the wider ward environment:

. . . it’s just trying to make sure ethically that we’re not doing things that are futile. Making sure that, you know, that we actually support and sometimes almost understand the impact of that decision not only for the patient or their family, but in the hospital environment actually how if there’s a horrible cardiac arrest how that impacts on the other patients, on the nursing staff, on the students . . .

Site 6, C06

Some doctors included the ReSPECT conversation itself in the balancing of harms and benefits, arguing that an incorrectly timed or framed conversation may burden patients.

The doctors who articulated a ‘harms and benefits’ perspective appeared to be guided by the ethics duty to protect patients from harm, characterised as ‘futile’ treatment, and to provide overall benefit (by balancing medical risks and benefits of possible treatments). The patient’s views and preferences were then sought in discussion of the pre-identified and selected options. However, if doctors were confident that one option was advisable, then patients were informed about the recommendation rather than included in decision-making. When patients’ and/or families’ wishes clashed with medical opinion, then most doctors chose the medical position, inferring limits on a patient’s autonomy, as one intensivist explained:

. . . if patients have unrealistic expectations we are not expected to provide that treatment just because the patients ask for it.

Site 4, C07

Some doctors described trying to achieve a compromise, whereby doctors denied futile treatment but explored with patients/families what could be done to respect the patients’/families’ values and preferences while not compromising the clinical duty to protect patients from harm. For example, a consultant described seeking consensus with patients from religious minority groups whose values did not necessarily match the values espoused by their doctors:

So there are some sections of the population who vehemently believe in preservation of life no matter what, at all costs, which is contrary to most of our views, but we have to acknowledge that and at least be seen to be doing everything we can to preserve life within what’s reasonable and acceptable.

Site 5, C03

A key aspect of negotiating between patient/family preferences and medical decisions was achieving a shared understanding to ensure that patients/families followed and agreed with medical logic. Most disagreements, doctors said, reflected misunderstandings:

. . . sometimes what happens is that the doctors, the nurses, the therapists understand the frailty and the futility of cardiac resuscitation, but actually the patient and the next of kin don’t.

Site 6, C06

To overcome misunderstandings, doctors checked patients’/families’ understanding and then provided an explanation to correct it. An acute medicine consultant said that when patients request CPR:

. . . I try to find out what is their understanding of resuscitation. Do they know what exactly we mean? Quite many times they think resuscitation means stopping treatment. So to clarify [to] them that we are not stopping any medical treatment.

Site 5, C04

Another acute medicine consultant spoke about considering patients’ backgrounds when explaining decisions:

. . . you’ve got to take that into account if you know that some patients will have different beliefs, different religious kind of backgrounds, and it’s about explaining to them that actually it’s not about withholding something and keeping it back but it’s about not giving the wrong treatment.

Site 5, C01

These doctors wanted to ensure that patients/families understood what was being recommended by the doctor and why. Doctors explicitly acknowledged patients’ values, wishes and concerns, and, in some cases, provided the possibility of modifying recommendations, therefore, respecting their autonomy, with the expectation or hope that patients/families would then agree with the recommendation.

There were some exceptions to doctors’ assertions that ReSPECT recommendations were ultimately medical. Doctors would respect a patient’s treatment refusal, although the doctors would sometimes try to convince a patient to reconsider, and this was not surprising given the legal requirement for a valid consent before treatment can be initiated. In palliative care contexts, the ReSPECT decision-making process focused more explicitly on patients’ wishes, preferences and values. A surgeon whose patient was diagnosed with inoperable stage IV cancer described the ‘main aim’ of the ReSPECT conversation as understanding ‘what she wants from the rest of her life, and how she wants to be treated’:

. . . as a doctor, it’s my duty to wish, to respect my patient decisions. Even if she comes to say to me she will like the full treatment, she wants the CPR, I, it’s my duty, I’m duty-bound, it’s, I’ll be happy to do it.

Site 6, C07

Similarly, in geriatrics wards, where there was substantial overlap with palliative care, doctors centred ReSPECT conversations on the patient’s wishes, framing the discussion through the lens of ‘What’s important to that patient right now?’ (site 3, C07), instead of limiting possibilities.

Doctors also said that they were more inclined to consider patients’/families’ wishes for treatment escalation in cases of uncertain prognosis or low likelihood of cardiac arrest. For example, this consultant contrasted futile and low-risk cases:

. . . if I had someone who was actively dying, we’re palliating, yet the family were adamant that he be resuscitated, I would be very firm in that situation that we were not resuscitating, that was my decision. For some, for a patient who’s otherwise well, is in for something minor and is going home tomorrow with no life limiting diagnoses, I would be more flexible.

Site 6, C02

Other doctors said that recording a ‘for CPR’ decision for patients unlikely to benefit from CPR was a form of showing goodwill to patients or their families. Speaking about an elderly patient who was undecided about his CPR wishes, a doctor said:

. . . but when it comes to CPR, to a certain extent, I think there’s a degree of showing that we’re doing what we can . . . so if it was his wish, and his wish was for us to try, I wouldn’t mind, you know, I wouldn’t not do it.

Site 4, C10

Some doctors talked about the importance of acknowledging the patient’s family’s emotional needs, particularly in the context of CPR recommendations.

You know, I have hundreds of patients over a year, they have one mum. And actually, if it’s important for them, I must, as [inaudible] part of the grieving process, it’s important for them to feel like everything had been done, even if maybe it wasn’t appropriate.

Site 3, C08

This acknowledgement was sometimes linked to concerns about future complaints or litigation. For example, a consultant described recording a ‘for CPR’ recommendation to avoid distressing the family and leading to complaints.

. . . what we don’t [want] is to leave behind people who are bitter and unhappy and very distressed that nobody was listening to them. Because it ultimately, it’s not the end of it, they usually start complaining to different institutions, to ombudsman, to the hospital, it lasts a long time and it brings no closure for them.

Site 4, C01

Clinicians perceive the ReSPECT form as bringing patients’ wishes to the foreground

Although most clinicians prioritised the medical perspective in the decision-making process, clinicians also perceived the ReSPECT process and form as foregrounding patients’ wishes, describing this as a key benefit. Some clinicians found the paper-based ReSPECT form helpful when engaging patients/families in discussions. For example, one clinician said that they used the form with patients whose prognosis was uncertain to involve them in decision-making, and this process proved useful even when patients did not have a definite preference to record:

But it, it allows, it’s to open the door to a, to a further conversation and allows them to go away and think about it and discuss with their family and to, and to talk about what they would like to do.

Site 2, C05

According to one clinician, the ReSPECT process allowed clinicians to facilitate patient autonomy by encouraging patients to think about what might be important to them in a future scenario.

Although most clinicians did not use the ReSPECT form during conversations with patients, several clinicians cited the form as helpful in prompting, guiding and structuring conversations with patients, expanding their remit beyond CPR decisions. This was especially the case for less experienced doctors, as one early-career doctor explained:

I think previously I, I didn’t know about ReSPECT forms, so it was always about resuscitation and ward-based ceiling of care.

Site 3, C03

Expanding conversations to include various treatment options, one clinician explained, allowed for a greater consideration of patients’ wishes:

. . . by having a ReSPECT conversation, we can better gauge what a patient would deem acceptable versus what they would say, ‘No, I don’t want this’.

Site 6, C06

Moreover, several clinicians said that the ReSPECT process, through eliciting patients’ wishes, enabled alternative understandings of quality of life, and that this led to changes in treatment recommendations.

Clinicians perceive the ReSPECT process as influencing treatment planning, although not all agree clinicians on how to use it in practice

Numerous clinicians said that the ReSPECT process broadened their thinking about treatment planning, prompting them to think about treatments other than CPR. This allowed clinicians to acknowledge patients’ complex needs:

. . . the DNACPR form didn’t say, didn’t have anything in between, if you like. So it was either not for resuscitation or for resuscitation, there was nothing else in between . . . but the ReSPECT form allows you still to choose something in the middle. It doesn’t have to be all for treatment or none for treatment.

Site 2, C03

Another consultant explained that this change in mindset led clinicians to construct treatment planning as provision, rather than denial:

And also I think it helps clinicians to understand that it’s not just a negative thing about what we don’t do. It’s a little bit about what we will do as well.

Site 5, C08

For some clinicians, the broadening of thinking about treatment planning translated into more detailed conversations with patients/families:

. . . the difference that ReSPECT has made from resuscitation forms is that I think we have, I have a lot more open discussions now regarding [what] people would and wouldn’t want . . . You know, so, coming into hospital or, you know, i.v. [intravenous] antibiotics, i.v. fluids, that sort of thing, all the way up to, you know, resuscitation, intubation, and things like that.

Site 4, C10

A change in mindset about the scope of treatment planning was closely linked with a change in clinicians’ understandings of patient/family involvement in treatment planning conversations. One consultant explained the contrast with DNACPR, which recorded a medical decision, sometimes without patient involvement:

. . . we’ve realised that actually that decision is ultimately ours, and we’re accountable for it, but it is wise and sensitive and probably appropriate to engage the family in that decision, and at least making that decision with their agreement. And then possibly, more recently, the, you know, the patient’s cultural views and the ways that they would want to be treated do now feed into that as well.

Site 5, C03

Although this consultant did not cite the ReSPECT form directly as leading to this change of practice, other doctors said that the form improved the quality of discussions:

But knowing that I’ve got to think through bits on the form, or needing, be needing to, to complete the form, I do try to, to broaden my approach, rather than just a simple question usually. Even if it’s only to ask in general terms, or, what people’s expectations would be if they got less well.

Site 6, C05

This suggests that the ReSPECT process encourages clinicians to consider the patient as a person, rather than focusing on a specific condition or event, emphasising benefit to the patient rather than avoiding harm from treatments.

Although some clinicians perceived the ReSPECT process as expanding the scope of their treatment planning conversations and recommendations, other clinicians described the ReSPECT process as similar to DNACPR:

But ReSPECT is, despite the good intentions, is just a different word for DNACPR . . . If I mention that I’m going to have a ReSPECT discussion for a patient who is having some elective operation, like a hernia operation, people will go, ‘Are you mad?’.

Site 5, C09

Many clinicians said that the ReSPECT process did not work as a universal form, and some clinicians avoided completing the form for patients who were for full active treatment. Without a ReSPECT form in place, the default understanding was that the patient was for all treatment:

. . . if the person is for a full active treatment . . . then we wouldn’t have a ReSPECT form in place.

Site 6, C01

In part, this reification of CPR decision-making reflected the form’s design. One middle grade-level doctor, who identified as an advocate of the ReSPECT process, said that:

. . . the ReSPECT form currently that we have is a, as I say, a black or white answer, it’s for or not for resuscitation. And so having that conversation becomes quite pivotal to the, to the rest of that conversation on the ReSPECT form.

Site 4, C10

Along similar lines, another clinician said:

But either for or not for resuscitation, that’s the only real decision that that form prompts me to do.

Site 2, C02

This conceptualisation of the ReSPECT form as DNACPR was shared across medical teams. One consultant explained that nurses equated the ReSPECT form with a DNACPR decision:

. . . sometimes the nurses were [inaudible], said, ‘Oh, a patient’s deteriorated, they don’t have a ReSPECT form’, because the ReSPECT form has replaced our resuscitation form.

Site 1, C06

Clinicians perceive the ReSPECT form as improving communication through structured and detailed documentation

Doctors noted that the ReSPECT form provides space to document the patient’s future care plan (as opposed to DNACPR forms):

. . . because planning for the future is a big part of the ReSPECT process, it’s not only about do not resuscitate decision, it’s about the, the planning for the future. It gives us the opportunity to, to discuss it with the families. And the, the place to document this discussion.

Site 4, C01

This form structure was especially important for palliative patients or those who required multiple treatment decisions. As one geriatrician explained, when speaking about a patient who asked to die at home:

I was able to write a very specific future care plan, which was ‘Avoid, relieve breathlessness, try and avoid coming back to hospital, make’’ you know, ‘slightly relax control of [condition]’, and those, those types of things.

Site 3, C07

In another case, a doctor explained that, although they used the ReSPECT form as a DNACPR and ceilings of treatment form in hospital, they also used the ReSPECT form to prompt ACP discussions between GPs and outpatients:

In outpatients I have used ReSPECT conversation as a trigger for advanced care planning, I don’t usually complete the advanced care planning myself in outpatients but I will prompt GPs and patients to have the conversation, or district nurses.

Site 2, C04

Several clinicians emphasised that the ReSPECT form did not change their conversational practices. In some cases, clinicians explained that this was because they already practised patient-centred communication:

I think as a geriatrician we have always done these and we’ve always done them well.

Site 6, C06

Other clinicians perceived the ReSPECT form as providing a new structure for documentation, without affecting communication practices:

I suppose that I would tend to fill out the form after having the discussions and more use it as a documentation process rather than having the form in front of me I suppose.

Site 6, C01

In some cases, clinicians said that the ReSPECT form has not changed their documentation practices, asserting that they saw no material difference between documenting the discussion in the notes or on the ReSPECT form:

. . . it’s always part of the plan for me, it’s always part of the plan. It was red form in [the] other hospital and previously we just used to write it in the notes, so it’s always part of the plan.

Site 2, C06

The ReSPECT process promotes quality and continuity of care in the hospital and across settings

Many clinicians noted that the ReSPECT process led to improvements in quality of care. For example, one consultant explained that the ReSPECT process reduced the harms associated with do not attempt resuscitation (DNAR) forms, particularly the mistaken withdrawal of treatment:

. . . so when I was a junior doctor and we had DNAR forms, one thing that always upset me was that if we had a patient on a DNAR form, some doctors assumed that meant do not treat, so they treat them with nothing . . . The ReSPECT form is much clearer, and the ReSPECT process is much clearer about this is not just . . . I think it, it has eased the confusion that it’s not just about CPR.

Site 5, C08

Another consultant said that, before the implementation of the ReSPECT form, treatment recommendations, even when recorded in detail, were not easily accessible, leading to harm in emergency situations:

. . . before we just used to write it all in the notes. But then there’s a risk of it being missed, you know. Beautiful two pages of prose written in the notes a week ago, then when the patient arrests no one necessarily knows that it’s there, or can’t find it. But having it as a ReSPECT form in the front of the notes, easily recognisable, easily found, and also on [the hospital’s digitised notes] there on every handover, that, that’s transformed it.

Site 4, C02

The benefits of recording decisions on the ReSPECT form extend beyond current admissions. One clinician explained that, when patients are re-admitted, the ReSPECT form allows for continuity of care:

. . . by having an instant look at it it can present with a good picture of the patient and the management if it’s done sensibly and properly. Then for the next admission, especially if they are recurrent patients, the same patients presenting again and again, then it’s such a timesaver for the next time for anybody else or ourselves.

Site 3, C01

With the ReSPECT form transferable across settings, several clinicians noted that the form promotes continuity of care between the hospital, ambulance crews and care or nursing homes. These clinicians include detailed instructions with other settings in mind:

. . . and that’s where ReSPECT comes in really helpfully, because then, the ambulance crew that, that responds to that 3 a.m. call can look at that and say, ‘Well, actually, this was the, this was the objective from that discussion’, and it would help their decision-making at the point . . .

Site 3, C07

Overall, clinicians perceived the ReSPECT form as supporting a more holistic approach to anticipatory decision-making that is inclusive of patients’ wishes and clinical needs.

Additional time and resources are needed to ensure that ReSPECT conversations are had with every patient who could benefit from them

Doctors cited time limitations as the main barrier to ReSPECT conversations. Compared with more urgent patient care, the ReSPECT process was said to be frequently deprioritised. Therefore, although doctors acknowledged the importance of having ReSPECT conversations, it is clear that doctors do not have ReSPECT conversations with every patient who could benefit from them. According to one geriatrician:

I don’t think this type of discussion takes place anywhere near enough, we’re working in a struggling health system that hasn’t got the resources to spend time having these discussions.

Site 3, C04

As a result, doctors tend to hold ReSPECT conversations with patients at risk of deterioration. The exception is one trust where doctors are prompted to have ReSPECT conversations with all patients, although doctors often resort to CPR-focused discussions. According to one, ‘[o]n a busy post-take round . . . it could be quite difficult to just go through absolutely every possible occurrence’ (site 6, C05). The other participating trusts do not have similar prompts and, as one junior doctor explained, ‘it’s just not really required and it’s quite a busy job so we just don’t have time’ (site 4, C09). Time limitations are inextricable from staff shortages, and some doctors explained that the ReSPECT vision cannot be realised with current staffing levels:

[It’s] a great idea, but it still has to be delivered by, you know, a limited number of people.

Site 5, C09

Owing to time limitations, doctors noted that the ReSPECT form cannot be used as a universal form, as one emergency medicine consultant explained:

I don’t think that we should do that for every patient in the emergency department because . . . it would be impractical. It would ground us, grind us to a halt if, if we had a resuscitation discussion with every patient.

Site 5, C08

Doctors’ reasons for holding ReSPECT conversations selectively rather than universally were not only practical, but also ethical. For example, a geriatrician explained why ReSPECT conversations are held with only a minority of patients:

. . . [i]f I have too much workload on any particular day I won’t be feeling comfortable to fill in ReSPECT form unless I know them pretty well. Because then I feel that I’m not doing it honestly or, or adequately if they are just very new patient and I don’t know much about them.

Site 3, C01

For this doctor, to have a hurried conversation shows lack of respect for the person whose treatment and care will be affected by the recommendations, and a hurried conversation could also lead to poorer care if considerations of treatment options are not adequately discussed.

Another geriatrician characterised ReSPECT conversations as emotionally laborious, saying that these conversations could not be held quickly and consecutively:

. . . they’re not conversations that should feel rushed and they, they should feel personally and it’s exploring the, the patient’s wishes and, and their family’s wishes. That does take time, I think that clinicians are human and very heavy discussions about ceilings of care and, and future wishes are, are draining.

Site 3, C04

Doctors said that trust policies placed additional barriers on the ReSPECT process. In particular, the requirement that consultants sign or countersign ReSPECT forms means that nurses and junior doctors, who are often the health-care professionals closest to the patients, either cannot or do not complete ReSPECT forms. Speaking about a patient who did not have a ReSPECT form, a junior doctor said that ‘she should have some sort of ceiling of care written down in her notes. But this ward is, kind of run by very junior doctors and we’re not always allowed to make those decisions’ (site 4, C06). Another junior doctor noted that unclear trust policies about countersigning made the ReSPECT process more complicated:

. . . all these rules floated around that if an FY1 [foundation year 1 doctor] was to do it, it would last for 24 hours, if a registrar would do it, it would last for 4 days, if a consultant does it then it would last forever, kind of weird rules that different wards didn’t agree with and some didn’t believe . . .

Site 4, C08

To overcome this, some consultants suggested that limitations should be lifted and all qualified staff be allowed to complete ReSPECT forms, although others cautioned that junior members of staff may not have sufficient experience and training. Balancing expertise and experience against time limitations remains a difficult ethics challenge.

Many doctors suggested that patients would benefit if ReSPECT conversations were held in primary care or outpatient settings:

. . . the real big change that’s needed is with those conversations we had earlier and not in extremis, in acute admissions, but in outpatients and in general practice so that it is not done at the time when things are, people are already very vulnerable and scared and are ill.

Site 3, C08

Although doctors tended to idealise the GP–patient relationship, describing it as imbued with a trust that patients do not feel towards hospital clinicians, some also cautioned that ReSPECT conversations held in community settings might not adequately address acute care scenarios, and would need to be reviewed on admission to hospital.

Improvements in the ReSPECT form, clinician education and transfer across settings are needed to facilitate higher-quality conversations and decision-making

Many doctors noted that aspects of the ReSPECT process conflict with clinical practice. Some doctors critiqued the ReSPECT form’s level of detail (considering it either too detailed or not detailed enough), the number of signatures required and the form’s lack of sufficient space for treatment recommendations. For example, a consultant who prepared a ReSPECT form as part of a patient’s discharge plan said ‘I was, you know, really having to struggle to get on with my text into the box at the bottom’ (site 6, C01).

Some doctors expressed reluctance to use the ReSPECT form in place of existing forms. One consultant explained that their ward’s communication form is superior:

And I would hate for the ReSPECT form to replace our communication form, because I feel that the ReSPECT form doesn’t have enough space to really document the, to keep a narrative of what’s been discussed.

Site 1, C06

Going further, another doctor said that the ReSPECT form is ‘just paperwork’ (site 6, C03) and that discussions are still recorded in the notes.

In most cases, however, doctors complete ReSPECT forms following discussions with patients/families. However, doctors overwhelmingly complete the forms for patients who are not for CPR. and most doctors explained that this is because ‘the default position, unless it has been explicitly discussed or documented, is that everyone is for CPR’ (site 5, C03). Some doctors worried that completing a ReSPECT form for patients who were for CPR would put patients at risk because medical teams perceived the ReSPECT form as equivalent to a DNACPR. One acute medicine doctor explained that they documented ‘for CPR’ decisions in the notes, rather than in a ReSPECT form, to avoid a catastrophic scenario, and another doctor explained:

. . . unless we can identify very quickly which patients are for resuscitation and which patients are not for resuscitation, putting in a ReSPECT form is going to create a lot of confusion. And I don’t want to be, I don’t want to be the person who makes a mistake . . .

Site 2, C01

As a result of this practice, doctors sometimes have repeated ReSPECT conversations with patients who are for CPR and have no documentation of these conversations.

Doctors noted that ReSPECT forms are often inadequately completed. For example, one geriatrician said ‘we have seen very much poor quality ReSPECT forms being done’, with doctors completing these forms ‘in 2, 3 minutes without doing justice’, thereby leading to forms that ‘are so general, they don’t feel [like an] individual plan for any individual patient’ (site 3, C01). Another doctor explained that, owing to the poor quality of many completed forms, ‘if you see a ReSPECT form you do wonder, you know, what’s the conversation that’s been had before. Because it’s not filled in very well most of the time’ (site 1, C06).

To overcome shortcomings in form completion, some doctors suggested changes in design. One consultant suggested that the form should feature ‘a short list of major interventions, you know, critical care, CPR and others to be specified’ (site 1, C04). Two junior doctors working on the same ward said that the ReSPECT form should include checkboxes for escalation decisions, such as ‘non-invasive ventilation, invasive ventilation, those kind of things’ (site 4, C09), to prevent an exclusive focus on CPR and increase clarity about future treatment, and ‘then that way all the parameters are thought about’ (site 4, C08).

Another design change suggested by doctors is the inclusion of uncertainty. Several doctors noted that they do not complete ReSPECT forms if patients/families disagree with medical advice or if patients/families are undecided about the treatment they want. Where the ReSPECT form has been digitised, doctors cannot complete the form without clicking on the CPR box, meaning that some ReSPECT conversations that are held are not documented.

Doctors also said that training should be provided to improve the quality of ReSPECT form completion. A geriatrician noted that gaps between specialties meant that some doctors complete the ReSPECT form as intended, whereas others focuse exclusively on CPR because ‘not everybody works in a holistic way’ (site 6, C06), necessitating additional education for clinical staff. An intensivist (site 5, C03) said that ‘quite a lot of resourcing, education and developing skills in clinical staff’ are needed to ensure that doctors havedReSPECT conversations early in a patient’s admission:

So education of, not just patients and the public, but also staff as well, and the confidence and skills to actually think about these issues and raise them with patients, and not just leave it to the intensivist who might be providing the last bit of resuscitation and then post-resuscitation care.

Site 5, C03

Despite the intentions of its developers, the ReSPECT form frequently does not transfer well across settings. Nurses often forget to provide the form to patients on discharge or systematically keep the form in the medical notes, nursing home staff sometimes misplace the form, and patients sometimes forget to bring the form with them to hospital. In hospitals where the ReSPECT form is digitised, doctors often do not provide patients with a paper form to take home, and this increases miscommunication across community and hospital settings, as one consultant noted:

We’ve got difficulties, particularly around the, the intersections of different sectors of healthcare. Primary care to secondary care, from secondary care to care homes or community. And it’s at those junctions in care where the, the form should flow with the patient.

Site 6, C05

To overcome this, some doctors suggested that an electronic system should replace the paper form. For example, a consultant said that ReSPECT ‘would probably need to be electronic because of course, you know, when the patient dials 999 for an ambulance, the one thing the ambulance men won’t remember to bring is the ReSPECT form that’s lying around gathering dust somewhere’ (site 1, C04).

Other doctors said that a national registry shared across NHS trusts and community care settings would be ideal, if unlikely.

ReSPECT form digitisation should be improved

The ReSPECT form had been digitised in two participating trusts. Although some doctors cited this as beneficial for record-keeping, other doctors felt that digitisation detracted from the ReSPECT process. In one trust, doctors said that the digitised ReSPECT form is unwieldy, leading to duplication of records, lack of patient interaction with the form and the recording of resuscitation status only, rather than detailed treatment plan recommendations. One consultant said that digitisation means that emergency medicine doctors avoid completing the form:

I know lots of consultant colleagues who complained about not being able to make the thing go green and sign it off, and I think that’s, I think anecdotally, certainly, it’d stop people completing it when they’re seeing people, particularly in the ED [emergency department], so it puts people off doing it on admission.

Site 6, C02

This consultant also said that the digitised ReSPECT system prompts doctors to complete a new form with every admission, making it an untenable option for outpatients and, therefore, ‘having a paper form in front of their . . . dialysis folder is by far the easiest solution’ (site 6, C02).

In one trust, digitisation has affected the transferability of the ReSPECT form across settings. As the digitised form cannot be printed, when patients are discharged, doctors have to copy the information onto a paper form, leading to duplication and often to no forms being provided to patients:

If anything, I think it’s a barrier to us generating the forms. They should be able to be just printed off . . . But I think that form’s a barrier, because you have to find the form and you have to fill it in.

Site 6, C05

Doctors were also concerned that the digitised form has lost the visual impact and meaning of the original paper form, and this is particularly the case with the patient preference scale, which is translated into percentages. Doctors noted that this scale is ambiguous and difficult to interpret in practice, As one consultant explained:

Because, well, to say 10% comfort care, 90% curative, or 20–80, or whatever, you know, I don’t know how people will really practically interpret that. And, you know, people’s interpretation will vary from person to person, I think. There’s not a dial that you can set 20–80 on the patient, or on the monitor, or on the infusion pump, or . . . you know, it’s, what does that translate to?

Site 4, C02

The solution to this ambiguity, one junior doctor learned, is to ignore the patient preference scale:

. . . initially when I’m like, asking my seniors, ‘What do I put for numbers?’ and I think everyone is just like, ‘It doesn’t really matter’, almost . . . I think everyone sort of tends towards comfort, but the numbers don’t really . . . We’re not really sure how to use the numbers basically.

Site 4, C05

Digitisation also means that patients cannot interact with the form during ReSPECT conversations:

And I think also the form being on a computer screen that’s, doesn’t, you know, it’s not easy, it’s not an easy thing for patients to interact with either.

Site 6, C02

The ReSPECT process needs to engage more carefully with patients’ lived realities

Many doctors noted that the ReSPECT form is less useful in emergency and acute medicine. The form, they said, was designed for patients who are well enough to engage with it meaningfully and not for acutely ill patients. As a result, doctors often do not use the patient preference scale. One consultant explained:

A frail patient in pain who’s had an acute admission, I think it’s entirely inappropriate to ask them to put a cross, because it’s meaningless, you’ll get them to put it one day, they’ll put it in a completely different place the other day. And it’s just not valid.

Site 2, C02

This approach extended beyond the patient preferences scale to include the ReSPECT conversation more broadly. Often, doctors take a persuasive approach or focus on resuscitation alone to avoid overwhelming acutely ill patients:

I think the challenge is the very sick person at the front door who feels lousy and doesn’t really have the energy to engage with a detailed conversation, but medically we still need to make some decisions about ceilings of care. And we end up being perhaps a little bit more directive than ReSPECT was originally designed for.

Site 2, C04

Doctors also noted that, although the ReSPECT process is useful in outpatient settings, the discussions that the ReSPECT process are meant to elicit – about quality of life and future preferences – re untenable in an acute setting. As one emergency medicine consultant explained:

I don’t think that anyone can, anyone who’s sick enough to be in a resus room and having a discussion about their care has the mental capacity to be making those judgements because I couldn’t do it now.

Site 5, C08

To overcome mismatches between form design and acute contexts, several doctors said that ReSPECT forms should be completed in primary care or outpatient settings, where patients are better able to speak about quality of life. 50

An additional concern raised was that the ReSPECT process and form might imply that treatment is available when medically impossible. For example, a consultant said that asking patients about their wishes may introduce false hope:

And I tell my colleagues, ‘Why do you ask them what they wish if you are not going to offer that?’ ’Cause then if you ask them and they say, ‘Yes’, you are, they expect you to do it. But it might not be the, from a medical point of view it might not be the appropriate thing to do.

Site 4, C07

Doctors were also concerned that patients/families might not be informed enough to make the right decisions. Doctors frequently said that many patients have never considered CPR, escalation of care and other treatment planning options before being admitted to hospital and, therefore, ReSPECT discussions are often limited:

As most of them end up being quite one-sided, it, I found that it ends up being less of a discussion that I would like, but more of a . . . ’cause normally, I found that most patients don’t really have clear thoughts about what they’d want beyond DNACPR.

Site 4, C08

In other cases, doctors worried that patients might express wishes for active treatment without knowing what this treatment entails:

I still have some concerns that does the average patient understand enough about these treatments to be able to make those decisions? And are they inherently more likely to want treatments that actually aren’t likely to be beneficial and could be burdensome?

Site 3, C05

Doctors said that the risk of making uninformed decisions should be mitigated during the ReSPECT conversation, but that not all doctors provide patients and their families with enough information:

. . . a slight danger of ReSPECT would be people only having half the discussion and saying that the patient wishes, for everything, it should be for everything rather than what’s medically appropriate for their treatment options.

Site 3, C04

To pre-empt situations where patients/families express uninformed preferences, doctors suggested that patients, and the public more broadly, should be educated about the ReSPECT process and the implications of particular treatments.

Sampling and recruitment

The six hospitals involved in the observational study also participated in the records review. Working with a pilot sample of 20 completed ReSPECT forms, a sampling frame, categorising records by ward, admission type (i.e. elective or emergency) and patient age, was developed to maximise the collection of diverse records.

Data collection methods

In each site, patient records (i.e. completed ReSPECT forms and related notes, if available) were collected within the first 2 years following ReSPECT implementation. Hospital research staff followed the sampling frame for selection of records and any deviation from this was discussed with the research team (e.g. if it was difficult to identify forms for younger patients or forms where the patient was ‘for CPR’). The collected records included ReSPECT forms versions 1.0 and 2.0. Despite some differences, the two versions were largely similar and the completed forms could be analysed using the same analysis tool. Data collection took place between August 2017 and April 2020, covering ReSPECT forms completed between July 2017 and January 2020 (i.e. prior to the COVID-19 pandemic affecting UK health care).

Analysis

Following an accountability for reasonableness (AFR) framework, which focuses on ethics decision-making,51 a qualitative evaluation tool was developed for a structured analysis of the quality of completed ReSPECT forms. AFR is an ethics framework that was originally developed for resource allocation decision-making. AFR focuses on the process of decision-making rather than on specific ethics principles, stipulating that decision-making should be transparent, based on reasons agreed as justifiable by the relevant stakeholders and open to review. 51 The process should be consistently applied to justify the ethics principle of equity. The tool’s development was inspired by an evaluation tool designed for a study on ethics decision-making in critical care admissions. 52 The development process involved several sessions with Anne-Marie Slowther, Claire A Hawkes, Frances Griffiths, Karin Eli and Zoe Fritz. The team established assessment guidelines, and the tool underwent ongoing testing and refinement. The final version of the tool assessed ReSPECT forms for the following:

1. Consistency. Have the forms been completed to the standard indicated by ReSPECT form guidelines?53

2. Transparency. Do the ReSPECT forms and relevant sections of the patient notes contain complete information about patients’ preferences, diagnoses, reasons for treatment recommendations, assessments of capacity (where relevant) and who participated in the ReSPECT discussion?

3. Ethics justifiability. Do the reasons for treatment recommendations indicate that clinicians considered patients’ preferences and clinical conditions, as well as the benefits and burdens of specific treatments?

All patient records (i.e. forms and patient notes, where available) were analysed by Karin Eli, with additional analysis provided by Claire A Hawkes, Frances Griffiths and Anne-Marie Slowther (with each scoring about one-third of records). Following comparison and discussion of the findings, the team achieved consensus on the key findings.

Ethics considerations

Data were collected with CAG approval (without individual consent). Patient and clinician identifiers were redacted prior to transfer to the study team. NHS researchers sent the pseudoanonymous records to the ReSPECT Evaluation Study Team using an encrypted NHS platform. Records were stored in password-protected files on a secure University of Warwick server.

Findings

Of the 162 completed forms, the analysis included 141 forms (paper forms, n = 121; digitised, n = 20; version 1.0, n = 87; version 2.0, n = 54). Twenty-one forms were excluded because of duplication (n = 2) or incomparability with the rest of the sample due to one trust’s incorporation of the ReSPECT form within its overall patient record system (n = 19). Twenty-eight records did not include patient notes.

Consistency

All forms included a CPR recommendation. Most forms (n = 87, 61.7%) included a recommendation regarding a focus on either life-sustaining treatment or symptom control.

Most forms included free-text treatment recommendations (n = 119, 84.4%). However, the level of specificity and detail varied substantially across forms. Specific treatments (e.g. non-invasive ventilation) were mentioned in less than half of the forms (n = 65, 46.1%), whereas some forms (n = 54, 38.3%) recorded general instructions (e.g. ‘for ward-based care’), which could be misinterpreted as they were site-specific instructions and did not clarify what treatments were and were not available on the wards.

Most forms (n = 97, 68.8%) recorded who was involved in the ReSPECT discussion. Where patients lacked capacity, some forms (n = 31, 54.3%) included families’ names or roles, whereas others (n = 26, 45.6%) included no information about ReSPECT discussion participants or only the names of the doctors involved.

Transparency

Although acute diagnosis was recorded in a minority of forms (n = 61, 43.2%), lists of chronic conditions were recorded in most forms (n = 120, 85.1%). Few forms included reasons for treatment recommendations (n = 13, 9.2%) and only a small minority of these forms stated that treatment recommendations were informed by patient preferences.

In most cases, ReSPECT discussions were mentioned in the notes (n = 91, 64.5%), but only a few forms (n = 11, 7.8%) included guidance on where to find ReSPECT discussion records in the notes. At times, there were mismatches between information recorded on the form and in the notes, such as when notes detailed treatment planning recommendations not included in the form.

No formal capacity assessments were recorded in the available notes for patients without capacity. Some clinical notes (17/38, 44.7%) included references to patients’ capacity, but most did not.

Only a minority of forms (n = 49, 34.8%) included an appropriate senior responsible clinician signature in the correct place on the form, and this may be because of a design fault in the form that led to confusion about where the clinician completing the form should sign if they were also the senior responsible clinician.

Ethics justifiability

Of the forms that included reasons for treatment recommendations (n = 13, 9.2%), most referred to the futility of treatment only (e.g. ‘CPR likely futile’). The weighing of the burdens and benefits of specific treatments was mentioned in only one form. Very few forms recorded how patient preferences or values had informed the recommendations.

Most forms (n = 122, 86.5%) included information about the patient’s capacity. Although 65 patients were recorded as having capacity, their preferences were recorded in just over half of the forms [n = 34 (52.3%) on the patient preferences scale; n = 17 (26.2%) in the free-text box]. For the 57 patients who were recorded as not having capacity, just over half (n = 31, 54.4%) of the forms recorded that a person close to the patient had been involved in the ReSPECT conversation.

Sampling and recruitment

We conducted focus groups with GPs, in areas served by five of the six hospitals that participated in the study’s other work packages. The sixth site’s focus group could not go ahead because of the COVID-19 pandemic.

To obtain a convenience sample, the study team sent information about the research to key contacts in areas served by the hospitals (including a GP partnership, a GP interest group, a health and care commissioning group, and a palliative care specialist with GP and other health-care professional contacts) that facilitated recruitment in their networks. Five focus groups of between 3 and 10 participants were conducted, with 27 GPs participating in total. Participants in three focus groups had used the ReSPECT process in their clinical practice (Table 1).

As the study progressed, we learned that it would be valuable to include other community health-care professionals to advance our understanding of the implementation of ReSPECT in the community. We conducted two individual interviews, with a paramedic and a community-based palliative care doctor (group ID: Int1), representing one of the hospital areas. These participants were recruited by a palliative care specialist and both participants used the ReSPECT process in their clinical practice.

All participants were paid £150 and were offered Continuing Professional Development certificates for their participation.

Data collection methods

All focus groups were facilitated by Karin Eli, four focus groups were co-facilitated by Claire A Hawkes and one focus group was co-facilitated by Anne-Marie Slowther. In two focus groups, facilitators were joined by a palliative care specialist from the local hospital and the palliative care specialist responded to participants’ locally specific questions about the ReSPECT process. At the start of each group, facilitators circulated a ReSPECT form and described the ReSPECT process. Although working in areas served by hospitals using the ReSPECT form, some GPs had not encountered the ReSPECT form prior to the focus group and these GPs’ discussions were informed by their experience of other forms, such as DNACPR and specialised local forms (note that we use the term ‘ReSPECT-type conversations’ to refer to these broader experiences). The two interviews were conducted by Karin Eli over Skype™ (Microsoft Corporation, Redmond, WA, USA). Both interviewees had seen patients with ReSPECT forms, and the palliative care doctor had conducted ReSPECT conversations.

Both focus groups and interviews explored participants’ experiences of the ReSPECT form (e.g. completing it themselves and seeing patients discharged from hospital with ReSPECT forms), how participants initiated ReSPECT-type conversations and participants’ experiences of ReSPECT recommendations being communicated across the primary care–secondary care interface.

Focus groups were conducted between April and November 2019 and lasted between 60 and 105 minutes. Interviews were conducted in March 2020 and lasted between 50 and 70 minutes. All focus groups and interviews were audio-recorded. Recordings were transcribed verbatim.

Analysis

Transcripts were analysed using inductive thematic analysis. 55 Caroline J Huxley closely coded the transcripts at the semantic level and identified candidate themes. These candidate themes were discussed and refined within the research team (KE, FG, A-MS and CAH), resulting in five themes.

Ethics considerations

Written informed consent was obtained from participants before each focus group discussion or interview. In the report, we have removed identifying details and have assigned all participants a group ID that cannot be traced back to them.

The ReSPECT form is an end-of-life care document

Participants expressed an implicit assumption that the ReSPECT form is an end-of-life care document for patients who are frail, under palliative care or in the final stages of a chronic illness. ReSPECT-type conversations are predominantly initiated by the GPs themselves, and are typically triggered by a deterioration in the patient’s health:

So there’s, like, a point where a patient’s condition turns that you can identify and then you begin to have a discussion.

FG4

Participants felt that it is important to identify and record patients’ wishes while they have capacity. The ReSPECT process is also initiated as part of routine care home admissions, and our paramedic participant said that ambulance crews mainly see ReSPECT forms with care home residents.

Participants often had a ‘hunch’ or ‘just knew’ that the timing was right to have a ReSPECT-type conversation, and participants would use verbal cues from the patient (e.g. references to recent experiences in hospital) as an opening for the conservation. Less commonly, patients themselves initiate ReSPECT-type conversations to express their wishes to decline CPR in an emergency. These patients are often healthy, and our participants questioned the morality of completing a ReSPECT-type form in these cases:

. . . some patients will want a DNACPR in place when there’s really nothing wrong with them . . . they just don’t like the idea of going through resuscitation. Then, that’s a whole other minefield . . . am I really doing the right thing for this person when they could have a really positive outcome potentially with treatment?

FG2

The ReSPECT process is best carried out in the community

Participants agreed that ReSPECT-type conversations can be carried out well within primary care. Conversations are often planned in advance, and so patients are prepared to discuss end-of-life care. Some GPs have lengthy pre-existing relationships with their patients, and these relationships make the conversation easier:

. . . because I’ve known them for so long . . . you can easily just say ‘While you’re here? I know we, we haven’t talked about this before and I think it is time that we are going to talk what’s going to happen in the future’.

FG5

Participants described ReSPECT discussions as an ongoing process that takes time to complete. As patients are not usually at immediate risk, there is more time for ReSPECT discussions to unfold in primary care than in hospital. Furthermore, community palliative care allows even more time for ReSPECT discussions than primary care, and the palliative care doctor noted that they can develop rapport with the patient without rushing the process.

Both our paramedic and palliative care doctor observed that, although community settings are ideal for ReSPECT discussions, GPs do not issue ReSPECT forms for every patient who needs one. The GPs described time and resource constraints that prevent them from initiating ReSPECT-type conversations. GPs reported that they feel constrained by the 10-minute consultation slot and are aware of a busy waiting room outside:

I think as doctors we find it, we don’t talk about it because we, it’s an add on to our normal consultation in 10 minutes.

FG1

Our participants reported feeling reluctant to hold sensitive ReSPECT-type conversations with patients they have not met before:

So some people you know really well, and, and it’s easy to, to bring things up and, and to chat to them, and be frank with them, whereas sometimes you’re in a situation where actually, you’re being asked to do it, or you think it’s appropriate to do it, and then you, but you’ve never met them before, and then, then it’s a lot more tricky.

FG3

A lack of experience and confidence in end-of-life planning also prevents GPs from initiating ReSPECT-type conversations.

Several participants suggested that it would be beneficial for health-care professionals other than doctors to initiate or be involved in the ReSPECT process:

. . . how many district nurses, or nurses fill in ReSPECT forms? Because arguably they . . . know all the patients far better than we do.

FG1

The paramedic thought that ambulance crews might talk about CPR more accurately and honestly than GPs because they have witnessed the physical and psychological impacts of CPR. However, participants noted that ReSPECT conversations can be especially difficult when patients do not agree with clinical recommendations, and dealing with such situations requires specialised training, which most nurses and paramedics have not received:

I don’t think that necessarily precludes [them] from doing it at all. I just think there is probably a fair amount of training I suppose that would need to take place to be able to do it.

INT1

Although GPs reported barriers to completing the ReSPECT process in primary care, GPs ultimately thought that it is the appropriate place to hold such conversations. All participants were critical of the hospital-issued ReSPECT forms they had seen, and thought that forms are being completed by inexperienced junior doctors in busy environments.

The ReSPECT process is an emotional process

Participants gauged how emotionally prepared a patient is to have a ReSPECT-type conversation by assessing their reaction when the topic is raised. If the patient reacs with alarm or withdrawal, then the participants would delay the conversation:

. . . tentative early attempts at discussing prognosis and decisions to escalate treatment . . . were distressing for [the patient] and the family. And so, we sort of shelved it and ended up having to go back to it a few weeks later when it was much easier to then complete it.

INT1

Many GPs include the patient’s family in these conversations, for emotional support and to ensure that everyone understands the plan.

A few GPs reported that they are emotionally unaffected by ReSPECT-type conversations because of professional experience. More frequently, GPs find it hard to maintain an emotional distance, particularly when they know the patient well. Emotional reactions can be affected by the patient’s reaction. If the patient reacts positively, then GPs feel that they had ‘done a good thing’ (FG3); however, distressed patients often leave GPs feeling ‘quite emotional [having] gone through that with them’ (FG3).

Conversations are driven by cultural understandings of death

Participants held implicit views of what a ‘good death’ was. 54 A good death was described as involving no CPR or invasive treatment and occurring at home or in a hospice, as this is what patients usually wanted for themselves. These understandings often underpinned ReSPECT conversations and the medical recommendations GPs recorded. However, some participants acknowledged that, in some non-Western cultures, a ‘good death’ means maintaining life for as long as possible, rather than focusing on quality of life. Nevertheless, clinical recommendations are based on the doctor’s understandings of a ‘good death’, and these sometimes conflict with the patient’s wishes. Such disagreements tend to focus on resuscitation. When patients/relative want CPR attempted and GPs feel that this was inappropriate, our participants try to nudge them in the direction they think appropriate:

[I] start talking about how many times during resuscitation that you break ribs . . . And they, they buy that and then they, they hear it when I say it, because quite a lot of those patients have known me for so long.

FG5

The palliative care doctor said that he/she records the patient’s wishes even when the patient’s wishes conflict with medical advice:

I would record the decision but in the knowledge that it’s probably not going to change what actually happens in practice.

INT1

General practitioners noted that patients’ families often feel ‘scared’ by terms such as ‘death’ or ‘dying’. Likewise, families feel that ,by engaging in the conversation, they are ‘condemning’ their relative. A few participants were keen to break down taboos and normalise discussions about end-of-life care, and these participants suggested holding ReSPECT-type discussions earlier (e.g. on first diagnosis or as part of a routine check-up).

There can be difficulties translating the ReSPECT process across care settings

Participants described situations in which their ReSPECT recommendations translated into care, for example when paramedics used the ReSPECT process to decide whether or not to transport a patient to hospital. However, participants also described situations when their ReSPECT recommendations were not transferred into care, and often this involved patients being admitted to hospital, despite a recorded preference for remaining at home. Structural reasons sometimes accounted for this, including a lack of hospice, home care service availability, care home staffing or resources.

Several GPs were aware that their lack of knowledge of specialist interventions could make their ReSPECT forms less meaningful in hospital, as their forms focus on preferences around hospital admission and resuscitation, and on treating chronic or terminal illness, rather than emergencies.

Some GPs had not seen hospital-issued ReSPECT forms even though their patients’ discharge letters mentioned the ReSPECT process. All participants supported an electronic version of the form that would be transferable between settings:

. . . so that all the different people providing care for a particular patient have got the . . . same kind of document that they can resort to in terms of palliative care and patient’s wishes.

FG4

For paramedics, digital sharing would enable the implementation of ReSPECT recommendations even when patients misplace the paper form. Without a digital record, paramedics spend time looking for a ReSPECT form if it seems that a patient should have one (e.g. if they see strong analgesics or oxygen tubing):

. . . it’s very reliant on . . . our own investigatory skills.

INT1

Participants critiqued hospital-issued ReSPECT forms and felt that the forms had limited value in community settings. Medical jargon commonly used in hospital-issued forms, such as ‘for ward-based care’ or ‘level 2 ceiling of care’, is not informative for community practitioners. In addition, hospital-issued forms tend to focus on specific treatments available in hospital, or are used as replacement DNACPR forms:

I feel like sometimes it is literally just, it’s like the same as the old DNACPR form . . . the resuscitation bit is like the tiniest little bit at the bottom. And the rest of that has kind of been left.

FG1

To improve transferability, several participants suggested including prompts on the ReSPECT form as to what information was required.

Engaging community health-care settings

Not all CCGs adopted the ReSPECT process, and none adopted the ReSPECT process simultaneously with the acute trusts. Owing to the variation in community adoption of the ReSPECT process, acute trusts’ engagement ranged from co-ordinating plans for implementation to raising awareness of the ReSPECT process in individual health-care organisations. One site mentioned that plans to engage with the public during implementation had not been realised. Implementers were concerned that (1) GPs were not aware that patients had a ReSPECT form following hospital admission, (2) GPs did not know that the ReSPECT form was patient-held and (3) the terminology used in hospital-issued forms did not transfer well to primary care or ambulance services.

Digitising the ReSPECT form effectively

Two trusts digitised the ReSPECT form. Digitisation improved ReSPECT form transferability, prompting, revising and auditing within hospitals, but also affected the implementation time frame and transferability across settings. Implementers identified the following challenges: (1) ensuring that digitised ReSPECT forms could be printed and given to patients, (2) ensuring version control and (3) engaging patients in ReSPECT conversations in the absence of paper forms.

To overcome the above challenges, implementation leads identified the following as effective strategies: (1) regular audits to support ReSPECT form quality assessments, progress towards implementation goals and training development; (2) mandatory training; (3) repeated/refresher training; (4) ReSPECT reminders (via hospital computer systems); and (5) continued revision of digitised ReSPECT forms based on clinician feedback.

Primary outcome

The proportion of resuscitation attempts abandoned because of futility is plotted by month, for ReSPECT hospitals and for other hospitals separately (Figure 3). The vertical line at December 2016 indicates the point at which hospitals first started implementing the ReSPECT process (see Figure 3). Prior to implementation of the process, the proportion of attempts abandoned because of futility was already falling rapidly, from approximately 9% in January 2015 to 4% in December 2016 in hospitals that went on to implement the ReSPECT process, and from 11% in January 2015 to 6% in December 2016 in other hospitals.

Post implementation, the proportion of CPR attempts abandoned because of futility fell a further 1% (to 3% in December 2019) in hospitals that implemented the ReSPECT process and by 2% (to 4% in December 2019) in other hospitals. Most of the reduction in proportion of CPR attempts abandoned because of futility observed during the entire study period (ReSPECT hospitals, 83%; other hospitals, 71%) occurred prior to the process being implemented. Predicted values from the ITS analyses models are also shown in Figure 3. Model fit appears reasonable but, nevertheless, there is considerable scatter about the regression line, with some relatively large differences between the predicted and actual values (e.g. at the start of the study in January 2015). For this reason, estimates of effect sizes from the ITS model (i.e. levels, slopes, differences in level and differences in slopes), reported in full in Table 3, may be unreliable.

FIGURE 3.

Interrupted time series analysis of the proportion of resuscitation attempts abandoned because of futility. Note that the vertical dotted line represents the date at which hospitals first implemented the ReSPECT process.

The difference in slope between the groups pre intervention was statistically significant (0.001, 95% CI 0.0002 to 0.0023; p = 0.02), but ReSPECT hospitals started from a lower level (on average, 0.04 per month less than in other hospitals), and this suggests that, prior to intervention, the two groups were not comparable. The proportion of resuscitation attempts abandoned because of futility did not change much in ReSPECT hospitals during the post-intervention period, but continued to fall, by 0.001 per month, in other hospitals. The difference in slope between the two groups post intervention was statistically significant (0.0005, 95% CI 0.0001 to 0.0009; p = 0.02), but there was no difference in the change in slope between the pre-and post-intervention periods in two groups (–0.0008, 95% CI –0.0019 to 0.0003; p = 0.17). Overall, therefore, there was no evidence that implementation of the ReSPECT process led to a reduction in the proportion of resuscitation attempts abandoned because of futility.

Secondary outcomes

Design

We carried out a retrospective observational study of routinely collected data from patient records and the NHS Safety Thermometer programme. 61 Participants were recruited from six acute hospitals that were using the ReSPECT process. To facilitate synthesis of findings, these were the same hospitals that participated in work package 1. Wards were selected by the research team, following discussion with participating hospital research teams, to ensure adequate numbers and coverage of the range of commonly found wards and clinical specialties found in district general hospitals.

Adult inpatients were included in the study unless they informed the research team they did not want their data used. Paediatric and neonatal patients, and adult day case admissions were excluded. The recruitment target was 3000 patients (i.e. at least 500 patients per site). Recruitment was co-ordinated to occur simultaneously with the routine monthly NHS Safety Thermometer audits. Five sites chose to collect data on a few wards each month and one site collected data on all participating wards on 1 day. Harm-free care was defined, in accordance with NHS Classic Safety Thermometer guidance:61

. . . the proportion of patients without any documented evidence of a new pressure ulcer category II–IV (developed at least 72 hours after admission to the site), harm from a fall in care in the last 72 hours, a new urinary infection (in patients with a urethral urinary catheter, which has developed since admission to this organisation) or new VTE [venous thromboembolism] (developed since admission to this organisation).

Contains public sector information licensed under the Open Government Licence v1.061

Patients whose care was not harm free per the above definitions were deemed to have experienced harm. We were unable to obtain any information regarding harms from medication (as proposed in the initial protocol), as this information is not routinely collected in all hospitals.

Participant’s opportunity to dissent from their records being used for research before and after data collection (opt-out and withdrawals)

As the intention was to assess all inpatients in selected wards at participating hospital sites on a single day, timed to coincide with each hospital’s NHS Safety Thermometer audit, we had permission from the CAG to collect data without obtaining individual informed consent, which was not feasible and would have introduced bias into the sample. Instead, leaflets were distributed to patients on the relevant wards in the days preceding data collection. The leaflets explained the study and gave details of how patients could dissent from data in their medical records being used for this research before data collection (referred to as opt-out from now on) or request withdrawal of their data from the study subsequently.

Overall approach

Our five research questions (listed in Objectives) were answered using a combination of basic descriptive statistics and regression analyses. Data on recruitment, completeness and content of ReSPECT forms, patient characteristics, harms and outcomes are presented using standard descriptive methods (see Objectives, research questions 1 and 5). To identify and quantify which patient characteristics and emergency care and treatment recommendations predict assignment to a clinician recommendation or DNACPR decision, or predict risk of harm, multiple regression techniques were used (see Objectives, research questions 2–5). Odds ratios (ORs) and risk ratios (as appropriate) and 95% CIs from basic (adjusted by recruitment site) and fully adjusted (multivariable) regression models are reported, with the former quantifying the effect of each characteristic on the outcome separately and the latter quantifying the independent effect (after adjusting for the other variables in the model). Clinician recommendation has four categories (i.e. ‘focus on symptom control’, ‘focus on life sustaining treatment’, ‘unclear’ and ‘not completed’). Therefore, for models to predict clinician recommendation, a multinomial regression was required (see Objectives, research question 2). Multinomial regression fits a series of pairwise models (subject to some overarching constraints) to the data, allowing comparison of each group to the reference category (‘focus on symptom control’ in this case). Clinicians would consider various patient characteristics to inform recommendations on the ReSPECT form. The choice of covariates for statistical modelling was based on these patient characteristics and the established knowledge of their likely effect on our selected outcomes. Variables considered for inclusion in the multivariable model were age at admission, sex, ethnicity, socioeconomic status, admission type, patient type, cognitive impairment, GO-FAR score,62 Charlson Comorbidity Index score64 (grouped by quartiles), McCabe Scale and involvement in making the plan. 63 We had planned to explore the effect of patient preference (i.e. section 3 of the ReSPECT form) on DNACPR and clinician recommendation; however, owing to this item being very poorly completed, patient involvement was used instead. Owing to collinearity between GO-FAR score, Charlson Comorbidity Index and McCabe Scale (because they are inter-related), only Charlson Comorbidity Index, which appeared to be the strongest predictor, was included in the multivariable models. Similarly, logistic regression with DNACPR (yes/no) as the dependent (y) variable was used to identify and quantify which patient characteristics predict assignment to DNACPR (see Objectives, research question 3), as well as estimate the effect of emergency care and treatment recommendations and involvement in making the plan on DNACPR decisions (see Objectives, research question 4). Patients were categorised as experiencing harm if they experienced at least one of the new harms (i.e. acquired during admission) listed in the NHS Safety Thermometer definition. Logistic regression with harm (yes/no) as the dependent (y) variable was, therefore, used to assess the effect of DNACPR decision on risk of harm (see Objectives, research question 5), while accounting for patient characteristics. Finally, as an ad hoc analysis, we use logistic regression with ReSPECT form (yes/no) as the dependent (y) variable to investigate whether or not patient characteristics, resuscitation status [DNACPR (yes/no)] and clinician recommendation influence chance of having a ReSPECT form. All regression models included recruitment site as a fixed effect (to adjust for any unobserved differences between hospitals) (see Appendix 1 for details of derived variables).

Statistical policies

Continuous variables that follow an approximately normal (or symmetric) distribution are summarised using means and standard deviations (SDs), and variables that were skewed are summarised using median and interquartile ranges (IQRs). Categorical data were summarised using frequencies and percentages. Distributional assumptions were assessed using the appropriate graphical methods (e.g. histograms and box plots). The significance level was set at 5%, with p-values of between 0.05 and 0.10 considered as weak evidence of an effect. There was no adjustment for multiple testing. The value of including categorical variables in the regression models was assessed using a linear test of trend (if ordinal) and a likelihood ratio test otherwise.

Sample size justification

For generalisability, we needed to ensure a reasonable spread of patients/decisions, and to model risk of harm we needed sufficient number of harmful events to have occurred. Based on data from one site’s experience with a similar ECTP, we expected 70% of patients to have had a ReSPECT decision, of whom 20% would have a DNACPR decision. Some 20–30% of patients were expected to have a no DNACPR decision, which, by default, means that they would receive resuscitation in the event of cardiac arrest. Assuming a ‘harm rate’ of 6.5% (based on 2014 NHS Safety Thermometer data61), enrolment of 3000 participants was expected to provide 200 incidences of harm for the risk modelling. The recruitment target for each site was, therefore, a minimum of 500 patients.

Recruitment

Recruitment began at the first site (i.e. site A) in October 2017 and was completed in April 2020. Data collection at the final site (i.e. site F) was cut short by the COVID-19 pandemic and so its recruitment total is less than the planned 500 patients. The numbers of participants eligible and recruited, by site, are given in Table 5. Data collection commenced between 8 and 11 months after initiation of the ReSPECT process at each site, and lasted 1–2 months (two sites) or 7–12 months (four sites).

The numbers of participants recruited by site, ward type and whether or not a ReSPECT form (full or screening) was in the patient notes are given in Table 6. Most (53%) participants were recruited from medical wards. The proportion of participants with a ReSPECT form ranged from 6% to 41% for full ReSPECT forms and from 6% to 88% for any ReSPECT forms.

Most (539/655, 82%) ReSPECT forms were completed after hospital admission. Among those forms completed after hospital admission, the majority (280/655, 43%) were completed within the first 48 hours of hospital admission (Figure 10). There was a very long tail, reflecting the fact that a small number of patients had been admitted for a long time before a ReSPECT recommendation was recorded.

FIGURE 10.

Histogram of time that ReSPECT recommendation was recorded following hospital admission.