CONSORT (2010) diagram of study participant flow, adapted for this study. 16 KIS-ACP = key information summary — advance (anticipatory) care plan. MDT = multidisciplinary team.
Relatively few eligible patients and recruitment delays meant 22 participants enrolled too late to reach the 48-week data point. Nine participants remained in the study and six completed the final questionnaires. The main feasibility outcome (a new or updated KIS-ACP) was measured at week 12 and assessed for all 46 participants.
Recruited patients were younger, having a median age of 65 years (range 59–71 years). Fewer recruited participants had pancreatic cancer (n = 12/46; 26%) because of its poorer prognosis and co-recruitment to another study. Table 1 shows the baseline characteristics of all screened patients (Supplementary Appendices S3.2–S3.4). Participants were keen to participate despite their diagnosis and debilitating symptoms or treatment side effects. Many wanted to share their experiences and improve care for future patients. Study retention was high in both groups with most withdrawals because of death or deteriorating health (Figure 1 and Supplementary Appendices S3.5 and S3.6). By 24 weeks, 30% (n = 14) of participants had died. At week 6, 64% (n = 28/44) returned their postal questionnaires. The researcher changed to collecting questionnaire responses by telephone in the final 8 months because of COVID-19 restrictions with a 66% (n = 21/32) overall completion rate at week 24.
Baseline characteristics of all screened patients
A total of 23 patients participated in semi-structured interviews, often with a carer. Ten case-linked GPs were interviewed. Table 2 provides details of the individual, dyad, and triad interviews. While completing questionnaires by telephone, participants often talked about their experiences. With verbal consent, 21 of these storied accounts were recorded as field notes, providing further rich data for integration with formal interview data.
Number of first, second, and dyad/triad interviews
Feasibility outcomes from patient clinical data found a new or updated KIS for 83% (n = 20/24) in the intervention group and 85% (n = 17/20) in the control group by week 12. An ACP plan was included in the KIS summaries of 45% (n = 9/20) of the intervention and 59% (n = 10/17) of the control group (Table 3 and Supplementary Appendices S3.7 and S3.8). The overall KIS-ACP quality was similar between groups, but almost half of the ACP plans were rated as poor by experienced clinicians. Medium-to high-quality ACP plans were mostly completed in the last 2–3 months of life. The letter helped 65% (n = 11/17) of participants answering the question about obtaining a GP appointment to get one. All 13 patients in the intervention group interviewed took the letter to their GP practice.
Outcomes analysis of GP urgent care records (KIS), ACP plan content, and quality
Relationships between GPs and patients, perceived busyness of the practice, and appointment systems had an impact on both groups. GPs thought the intervention letter a useful prompt to consider earlier engagement with patients. Documented future care plans were valuable but required a difficult, delicate balance between encouraging patients to engage in challenging conversations and respecting individual wishes.
Attitudes and perceptions of individual patients, carers, and GPs determined subsequent ACP outcomes and the interviews revealed four styles of care planning behaviour:
Early patient planners were uncommon and most had worked in health care. Many participants had not engaged in ACP by second interview. From 10 GPs interviewed, three intervention and one control group GP were early planners. Interactions between people with different planning styles emerged as more important than the intervention in determining outcomes; shown here in the Results and in Supplementary Appendix S4.
‘Our GP took the bull by the horns right from the start. She asked me about end of life and if I wanted resuscitated as soon as I was diagnosed, and I said no.’
(Control group patient [23P], second interview, 6 February 2020)
‘Some patients find it very difficult to engage with us as GPs or our primary health care team if they are very much fixated on what can oncology offer. Sometimes, we as GPs feel that patients actually haven’t got to that stage. There is a small group who cling on for that last bit of hope. So, that is led by the patient, unfortunately.’
(GP interview for control patient 23P, 6 February 2020)
‘It’s hard to say what he said. Something about this is about what you want towards the end of your life. But we’re just no feeling at that stage yet. I’ll sort of cross that bridge when I come to it.’
(Intervention group patient [43P], first interview, 10 April 2020)
‘He has a KIS, yes. We discussed ACP and he was meant to get back to me about it, but he hasn’t yet. And that was when all the COVID stuff came up, so that’s been disrupted.’
(GP interview for intervention patient 43P, 26 June 2020)
‘I handed it into the receptionist. And I’ll be honest with you, I did’nae [didn’t] really need the doctor, so I’ve never really bothered him. He’s [GP] never been in touch about the cancer. Looking ahead with the cancer, I’m nae [not] worried. Mine’s no [not] malignant, and it’s no [not] spread. If I get another 10 years, I’ll be quite happy.’
(Intervention group patient [26P], first interview, 6 December 2019)
‘I’ve never given that a thought. Well, I’m telling a lie there; I have thought about it and basically, I wouldn’t want resuscitated. I wouldn’t mind being asked that question, but I’ve never been asked. I feel that if my quality of life was over, I’d rather just be.’
(26P, second interview, 3 March 2020)
‘The KIS has been activated but it simply says a new diagnosis of atrial fibrillation. We don’t have a standard policy on palliative care. We are expected to make all of them a KIS with a palliative care summary. In reality, we have a lot of patients who you could call palliative, but who are very well and have a full, active, unrestricted life.’
(GP interview for intervention patient 26P, 3 March 2020)
Attitudes towards ACP and strongly held beliefs about its association with cardiopulmonary resuscitation and planning for the end of life conflicted with living well in the present, maintaining normality, and perceived harms of planning ahead too soon. Public and professional information about ACP being an ongoing process of preparing for the future well before the last phase of life had little impact on prevailing social norms about fighting cancer and maintaining hope. With few exceptions, ACP was reactive and triggered by changes in the patient’s health or cancer treatment plan. Pressure of time and the drive to complete resuscitation forms (DNACPR, do not attempt cardiopulmonary resuscitation) affected perceived behavioural control among GPs. Figure 2 illustrates factors underlying these ACP behaviours.
Theory of planned behaviour applied to advance (anticipatory) care planning.
Patients in both groups maintained a good quality of life despite their poor prognosis and the impact of treatment side effects. Baseline EuroQol EQ-5D-5L had a median Health Index score of 0.72 (Q1–Q3, 0.66–0.81) and visual analogue scale (VAS, 0–100) median score of 75 (Q1–Q3, 65.0–85.0). Scores remained stable to week 24 where the median Health Index was 0.71 (Q1–Q3, 0.55–0.77) and VAS score 70 (Q1–Q3, 60.0–85.0) (Supplementary Appendix S5.1).
Patients recorded high scores on five ICECAP-SCM wellbeing measures up to 24 weeks, shown as a range of scores: ‘Having a say most of the time’ (88%–100%); ‘Being with people who care about you most of the time’ (80%–97%); ‘Emotional suffering sometimes or rarely’ (85%–100%); ‘Able to maintain dignity most of the time’ (92%–100%); and ‘Being supported most of the time’ (93%–100%).
Scores for the ICECAP-SCM domain ‘Experiencing physical discomfort often or always’ (28%–50%) suggested most patients suffered increasing physical symptoms over time (Supplementary Appendix S5.2).
Interviewed patients shared experiences of living with cancer, and felt well supported with cancer treatment and involved in all aspects of decision making reflected in CollaboRATE scores from both groups. Patients recorded a top score for ‘Every effort was made to involve me’: baseline (65%), week 6 (59%), week 12 (48%), and week 24 (59%) (Supplementary Appendix S5.3). Trying to maintain normality and balance being pain free while minimising analgesic use and side effects were important considerations: ‘It’s amazing because these are all pretty serious things and yet I’ve felt good and I’ve felt healthy enough and I’m eating well. I feel quite positive although there are the odd days where I feel a bit anxious and things.’ (Control group patient [42P], fieldnotes, 31 July 2020) ‘Well, if I could just get to the bottom of this back pain that makes me feel sick, I’d be absolutely fine. I can take the immediate release [opioid] any time I feel it coming on. It’s trial and error.’ (Intervention group patient [38P], second interview, 23 July 2020)
Service use by both groups appeared broadly similar. Patients in the intervention group spent more days in hospital (median 14.0 days [Q1–Q3, 5.0–24.0] versus 6.0 days [Q1–Q3, 3.0–11.0]). Home deaths appeared higher in the intervention group (53% versus 29%) although numbers were small. Median time from multidisciplinary cancer meeting to first GP contact was shorter in the intervention group but there was substantial overlap between groups indicating diversity of experiences among participants (7.3 weeks [Q1–Q3, 3.6–8.9] versus 13.1 weeks [Q1–Q3, 2.6–19.4]). GP visits or contacts were similar; median 8.0 (Q1–Q3, 3.0–11.0) through to the last week of life (Supplementary Appendices S5.4 and S5.5).
Cancer treatment stopped around 3 months before death in both groups. Of 46 participants, 37% (n = 17) received specialist palliative care, with later referral in the intervention group (26.4 weeks [Q1–Q3, 9.1–48.4] versus 11.6 weeks [Q1–Q3, 3.7–22.4]). Other time-to-event data were similar for all participants. The median survival from cancer review meeting was 30.6 weeks (Q1–Q3, 17.0–42.9), and a KIS was started or updated after a median of 18.6 weeks (Q1–Q3, 7.6–42.9) (Supplementary Appendices S5.4 and S5.5). A total of 29 (63%) patients died during the study, all from cancer or its complications (Supplementary Appendix S5.4). No psychological or clinical harms were identified.
This study found patients with incurable oesophageal, gastric, or pancreatic cancers starting palliative oncology treatment can be recruited to an ACP trial in primary care with good retention rates even though one-third died within 6 months. An intervention letter from oncology was acceptable to patients but did not trigger more ACP discussions leading to documentation of a clinically relevant ACP. The findings of this study offer insights into how people navigate incurable cancer and cancer treatment by maintaining normality; reflected in high quality-of-life scores and personal accounts. Interview dyads/triads illuminated an evolving dialogue between patients, family carers, and GPs about care planning as the patient’s health deteriorated. Where care planning styles aligned or GPs could broach thinking about the future linked to clinical events, meaningful conversations often followed.
This study has several strengths. This feasibility trial successfully evaluated an ACP intervention designed with local cancer clinicians, GPs, and public–patient representatives, and its implementation in routine clinical practice. 30 Retention was good, with death or deterioration the main reasons for withdrawal. Validated questionnaires with good response rates measured quality of life and decision making about palliative cancer treatments. Health service use and time-to-event data mapped care pathways for a rapidly deteriorating patient group. Qualitative interview data generated in parallel were integrated with quantitative data for reporting. 31 Longitudinal, multi-perspective interviewing was used to generate rich data about underlying attitudes, beliefs, social norms, and external factors making earlier ACP discussions in primary care possible but complex and individual. 32
There were also study limitations. Many screened patients were not fit for cancer treatment or opted for palliative care. This slowed recruitment and meant follow-up to week 48 was not possible for participants joining the study later on. COVID-19 exacerbated difficulties recruiting GPs and obtaining primary care service use data. The pandemic may have had an impact on KIS completion and ACP quality through pressure to complete more plans that directed care in the event of COVID-19 infection. Excluding the final ICECAP-SCM wellbeing question on PPI advice prevented summative scoring across all domains. This study took place in a single cancer centre serving two Scottish health boards and participants did not represent ethnically diverse backgrounds, limiting generalisability.
GPs have an important, valued role in providing care coordination for people with incurable cancers, along with general palliative care support, as a patient deteriorates. 13 , 33 In this study, such care was evident particularly in the final phase of illness. Patients appreciated their cancer care and felt involved in treatment decision making but collaborative approaches to ACP between oncology services and primary care were lacking. Systematic reviews of ACP implementation studies have found that effective ACP programmes focus on individualised discussions over time and target multiple stakeholders to deliver integrated ACP. 34 ACP intervention studies in primary care are underdeveloped compared with other settings and lack cohesive structures. 35
Social norms associating ACP with death and dying, and fears about consequences of discussing them, were almost universal among participants. Research exploring end-of-life communication within families has shown how such normative values create multiple barriers to open dialogue including belief in positive thinking and protective buffering. 36 Many people in the current study exhibited fluid and oscillating ‘dual dialogue’ or double awareness; the capacity of individuals to live as normally as possible while also being aware of the inevitability of death. 37 , 38 This concept underpins successful early palliative care interventions in oncology out-patient services that support patients and family carers to gradually engage and cope with both living life well and acknowledging dying. 12 , 37
The COVID-19 pandemic highlighted harms associated with policy drives to record advance treatment plans, including cardiopulmonary resuscitation status, at the expense of meaningful conversations about what matters to people at risk of deteriorating health. 39 Shifting professional and public understanding away from ‘planning for dying’ towards broader, values-based concepts of ACP in the UK and internationally is increasingly accepted as best practice. 1 , 40 , 41 Patient–public information and education about ACP is becoming more widely available and aims to normalise ACP by empowering people to have these important conversations with their family, friends, and clinicians. 42 – 44 Primary care teams can be supported better to embed ACP in routine care, and start earlier conversations about serious illness, coping with uncertainty, and preparing for future changes in a person’s health by using resources developed by primary care leads. 35 , 45 Electronic care coordination systems for ACP in primary and secondary care are being developed to be accessible, readily updated by all key professionals, and clinically relevant. 46
Progression to a full trial of early ACP in primary care with people with incurable cancers is feasible in terms of patient recruitment and retention, and this study suggested a 24-week intervention may be sufficient. Longer follow-up remains valuable for secondary outcomes such as quality of life and health service use. People living with the challenges of incurable cancer and palliative oncology treatments were keen to participate in research and improve care for others. Timing and acceptability of ACP depended on perceptions and attitudes of patients, carers, and GPs. Understanding these behaviours is already informing future ACP research and clinical practice through the 2021 national toolkit for ACP in Scotland and development of a new digital platform that addresses current limitations of the KIS system. Primary care has a central role in ACP but collaborative models with oncology merit investigation. Public engagement campaigns should continue to promote wider discussions about future care and highlight benefits of planning for changes in health to help patients and carers navigate health and care crises. Values-based health care prioritises talking about what matters to a person; reflected in recent professional education on meaningful conversations about ACP.
The study team acknowledge the contribution from their patients with cancer, their families, GPs, cancer nurse specialists, and oncology clinicians. Their willingness to take part and generosity of time helped to support this study despite the onset of the COVID-19 pandemic. The patient–public involvement group, clinicians, and academics on the project steering group committee provided invaluable insights and contributions. Nicole Brun provided administrative support.
This study was funded by Macmillan Cancer Support (reference: 6488086). Christopher J Weir was also supported in this work by NHS Lothian via Edinburgh Clinical Trials Unit (ClinicalTrials.gov Identifier: NCT03719716).
Ethical approval was obtained from the West of Scotland Research Ethics Committee 4 (Integrated Research Application System reference: 18/WS/1035; project ID: 243938). Local research and development approvals for the two health boards and Caldicott Guardian approval to access health service use data were obtained.
All datasets are archived at the University of Edinburgh and available from the authors on reasonable request. The full statistical analysis report and additional qualitative data are provided in the Supplementary Data.
Freely submitted; externally peer reviewed.
The authors have declared no competing interests.
Contribute and read comments about this article: bjgp.org/letters