Lessons Learned From the Development and Implementation of a Shared Decision-Making Process for Severe Stroke Using a Digital Communication Platform

Akila Visvanathan, PhD; Sarah Morton, PhD; Allan Macraild, MsC; Polly Black, MsC; Sophie Gilbert, MsC; Mark Barber, MD; Martin Dennis, MD; Richard O’Brien, MD; Gillian Mead, MD

Stroke is the second leading cause of death worldewide.¹ Approximately 50% of people who have a stroke are left with ongoing disability.² Stroke symptoms range from mild neurological deficits that resolve quickly to severe, persistent life-threatening deficits. After severe stroke, treatment decisions may determine survival with disability or death.³

Because stroke occurs suddenly and unexpectedly, patients and their families and caregivers are often unprepared to make treatment decisions. Patients often lack decision-making capacity because of aphasia, cognitive impairment, or impaired consciousness.^4-6 Families may be in a state of shock, and prognosis for recovery is often uncertain.⁷

Shared decision-making (SDM) is important because the patient’s involvement in treatment decision-making is a fundamental right. Patients and their families generally want to take an active role in decision-making,⁸ but this does not always happen after stroke.^9,10 The quality of SDM and the patient’s quality of life after stroke are key priorities for future research in severe stroke.¹¹

To our knowledge, only one trial has evaluated SDM among patients after severe stroke. This trial studied the feasibility of a paper-based decision aid in a US neurological intensive care unit with 41 patients with severe acute brain injury and stroke and 66 surrogate decision makers. The study reported that the decision aid was feasible and well received.^12,13

To address the paucity of research on SDM after stroke, a team of health care professionals, in collaboration with the National Voices Group and Chest Heart and Stroke Scotland (CHSS), developed a digital communication platform called Tailored Talks. The platform uses PowerPoint slides to share customized information with patients and their families about stroke effects, prognosis, and treatment outcomes.¹⁴

The primary aim of this study was to co-produce and embed an SDM process for severe stroke into the stroke care services at one teaching hospital site (The Royal Infirmary of Edinburgh, Scotland), using Tailored Talks as the information source. Our secondary aims were (1) to evaluate whether the SDM process was effectively implemented; (2) explore whether it was associated with changes in processes and outcomes (ie, death, discharge destination, use of feeding tubes); (3) evaluate the views of patients, families, and staff working within the stroke service about the quality of SDM both before and after implementation of the SDM process; and (4) explore whether patients and families’ preferred outcome (ie, death/severe disability/neither) at baseline matched the actual outcome at 6 months.

Ethical approval was given by Scotland A Research Ethics Committee (21/SS/0044). We used mixed methods: (a) coproduction (months 1 to 4); (b) implementation (month 6 onward); (c) audit (months 1-12); (d) questionnaires (months 3-9); and (e) qualitative interviews (months 6-12) with patients and relatives, and a focus group with staff.

Coproduction

Coproduction is a collaborative research approach that involves multiple stakeholders underpinned by three principles: (1) actively engaging participants in a structured, participatory approach; (2) ensuring all participant voices are heard, evaluated, and appropriately acted on, and; (3) encouraging all participants to actively contribute to the development of the SDM process.

Our coproduction group included 13 participants who were recruited through stroke charities and professional networks. The participants were stroke survivors, relatives of stroke survivors, and stroke care professionals from a range of disciplines and levels of experience.

Participants were invited to one of two introductory workshops, which were followed by four coproduction workshops (hosted between January 18, 2022, and May 24, 2022) facilitated by at least two researchers (SM, AV, and/or AM). Each workshop was about 1 hour long and was hosted online (due to COVID-19 restrictions) using NHS Scotland National Video Conferencing service. Attendance decreased over the workshops, with 11 participants at the first workshop, eight in the second, seven in the third, and six in the fourth.

Each workshop covered four key topics: (1) discussing study aims; (2) delivering Tailored Talks in practice; (3) eliciting family and patients views; and (4) implementing the SDM process. Participants were invited to evaluate different intervention functions (eg, education, persuasion, incentivization, coercion, training, restriction, environmental restructuring, modeling, and enablement) in their appraisal of Tailored Talks and its role in an SDM process. The APEASE (Acceptability, Practicability, Effectiveness, Affordability, Side-effects and Equity) criteria were considered.¹⁵ The “Equity” criterion was not applied during the coproduction process since it is intended to appraise equity of the intervention in practice, rather than in theory. Workshops were recorded using an encrypted audio recorder and transcribed verbatim. Transcripts were imported into NVivo v11 for thematic analysis and coded by two researchers (AV, SM) before each workshop, allowing for an iterative approach that informed the discussions at subsequent workshops.

Audit

Between February 1, 2022, and January 31, 2023 (months 0 to 12), we extracted data from the Scottish Stroke Care Audit on death, place of discharge (ie, to institutional care), and the use of feeding tubes (nasogastric and percutaneous gastrostomy) for all patients with acute stroke seen in our hospital. Clinical staff agreed to record the National Institute of Health Stroke Scale (NIHSS) for all patients to identify those with severe strokes (NIHSS ≥ 15). The audit coordinator extracted information from medical records about the use of Tailored Talks. We used chi-square tests to compare differences in tube-feeding/death and discharge to institutional care before and after implementation.

Questionnaires

The research team identified potential participants (acute severe stroke, NIHSS ≥ 15) 3 months before implementation of the SDM process (May 1, 2022) and for 3 months afterward. If clinical staff had not administered the NIHSS to a patient who had a clinically severe stroke, GEM calculated the NIHSS retrospectively from information in the patient’s medical records. Patients with decision-making capacity and or their next of kin if they did not have capacity were approached for the study.

Research nurses administered the four-item SURE test, which is a short version of the decisional conflict scale that comprises four yes or no questions,¹⁶ and the three-item CollaboRATE measure to assess the perception of being informed and involved in decision‐making steps.¹⁷ The research nurses administered these assessments face-to-face or by telephone at baseline and at weeks 2, 4, and 8. These intervals were chosen based on our experience as these  are time points when key decisions about life-sustaining treatments and advance care plans are often made.

At baseline, research nurses also administered the simplified modified Rankin Scale questionnaire (smRSq)¹⁸ and asked two open-ended questions:

1. If your (or your loved one’s) stroke was so severe that you (they) could no longer look after themselves and require care in a nursing home, which would you (or your loved one) prefer?
   • ​​​​​​​Dying comfortably from the stroke in hospital
   • Dying at home after discharge for palliative care; or
   • Surviving with disability but needing long-term care in a nursing home?
2. As you (or your loved one) are now, which would you prefer?
   • ​​​​​​​Dying comfortably from the stroke in hospital;
   • Dying at home after a discharge for palliative care; or
   • Surviving with disability but needing long-term care in a nursing home?

At 6 months, we obtained data on the actual outcome (ie, death or discharge to institutional care), completed the smRSq, and asked about the patient’s specific abilities (ie, if the patient could walk, talk, and eat normally [yes or no responses]), and whether the patient had anxiety or depression, as assessed by the 5-level EuroQOL-5D. A follow-up phone call was completed at 6 months.

Qualitative interviews and focus group about the SDM process

AV conducted telephone interviews between November 16, 2022, and November 25, 2022. Guided by the Consolidated Criteria for Reporting Qualitative research,¹⁹ two researchers (AV and SM) independently coded transcripts using NVivo and performed thematic analysis.

We tried to engage staff through posters and discussions. Individual meetings were led by GEM; AV and AM conducted a focus group on December 6, 2022, with staff who were recruited in response to invitational posters displayed in staff rooms. The discussion was audio-recorded, transcribed, coded using NVivo, and analyzed (thematic approach) by AV.

Coproduction

The feedback obtained from the four workshops, ideas for implementation, and which aspects of the implementation plan were put into practice are shown in Table 1.

The implementation plan was registered with our department’s quality improvement lead before the official implementation date of August 1, 2022. Preparatory training in Tailored Talks was provided to staff before implementation.

Mapping of the feedback according to APEASE criteria are shown in Table 1 (appendix). Project “champions” and dedicated iPads for the project were recommended by our coproduction group, but we were unable to provide either.

Audit

From February 1, 2022, to January 31, 2023, 1020 patients (502 pre- and 518 post-implementation of the SDM process) with a diagnosis of acute stroke were admitted; the mean age of patients was 73 (SD ± 15), and 496 (48.6%) were female. We used an iterative quality improvement methodology to increase the proportion of patients with a documented NIHSS, but improvements were not sustained (Figure). The overall proportion of patients with an NIHSS assessed at admission by the clinical team was 581 (57%); of these, 143 (24.6%) had an NIHSS of ≥ 15. For the entire group, there was no difference in tube-feeding during admission, death, or discharge to institutional care at 6 months before and after implementation (Table 2). The low rate of discharge to institutional care from our ward was attributed to the referral of disabled patients to other hospitals for rehabilitation. We noted, however, a statistically significant reduction in death at 6 months for severe strokes between the pre- and post-implementation phase. It is unclear whether this is a chance finding.

Questionnaire data

Between May 1 and October 31, 2022, research nurses identified 78 potentially eligible patients by discussion with the clinical staff; of these, 37 had an NIHSS < 15; five died before recruitment; five had no capacity, no next of kin, or were not proficient in English; four declined; two were moved to another hospital or nursing home before recruitment; and four could not be reviewed after initial contact.

Of the 21 patients (14 women, 7 men) who were recruited, the mean age was 80 years (range, 41-95), and one patient was lost to follow-up before any assessments were conducted. Of the remaining 20 patients, surrogate responses were obtained from next of kin or family for 18 patients, and only two patients answered the questions for themselves. Median NIHSS was 23 (range, 15-34).²⁰ The preferred outcome at baseline is shown in Table 3. At 6 months, 14 patients had died, three were in a nursing home, and three were lost to follow-up.

Of the 14 patients who died, only one had stated at baseline their preference to survive as they were at the time, rather than die in a nursing home; all 14 patients said that they would rather die than live with a severe disability.

Table 3 summarizes the patients’ responses at the time of consent with respect to the two questions asked. Table 4 shows the CollaboRATE and SURE responses. We cannot draw any conclusions about participant perceptions before and after implementation because of the small sample size, or about functional status at 6 months.

Qualitative interviews

Five participants were interviewed (1 patient, aged 73, and 4 relatives of deceased patients [ages of deceased patients, 89, 83, 63, and 91]). The full quotes are in Supplementary Table 1. Three main themes were drawn from these interviews:  

1. Experience of stroke and stroke care. Some participants were complimentary of the care that they received on the ward. Value was placed on being shown brain scans and having the situation explained. Some participants could not remember what had happened when they were in hospital. Patients reported the shock of having a stroke and the fear of recurrent stroke.
2. Diagnosis and discussions about stroke and treatment and involvement in decision-making. Participants had varied experiences with discussions about stroke diagnosis, prognosis, and management options. Some felt unable to make choices or ask questions about treatment due to shock of the diagnosis, illness, or lack of awareness that choices existed.
3. Provision of information. Four participants valued information from health care professionals and found viewing brain scans was helpful for understanding their prognosis. Participants reported varying needs for the amount and type of information (eg, paper leaflets, online sources). Some participants felt the information should be given after the initial shock of stroke had lessened.

Tailored Talks as a mode of information provision

One participant who received Tailored Talks reported she had been shocked to hear about the prognosis so early after stroke. The participants who did not receive Tailored Talks felt that it might have been useful.

Focus group of health care professionals

Five participants were initially recruited, but only two could attend due to clinical service pressures. The full quotes are in Supplementary Table 2. There were two main themes:

1. Experience with Tailored Talks and its use. Participants found it to be a good source of information and a useful educational resource. Participants had used it for “low stake conversations” (eg, medication discussions) but not for “high stakes” discussions (eg, tube-feeding or end-of life-care).
2. Barriers to Tailored Talks use. Reported barriers included insufficient time on the busy ward, materials were “too medical” and not “patient-friendly,” loss of eye contact and “human touch,” and lack of perceived consultant leadership.

Despite using coproduction methodology to develop a new SDM process and implementation plan, our audit of 1000 patients suggests that the process was not effectively implemented and there were no changes in tube-feeding or death/institutional care at 6 months for the entire group. Our SDM process incorporated all the key elements recommended by the American Heart Association for cardiovascular SDM, except for the provision of “decision coaches.”21 Below we discuss some of the challenges we faced and potential reasons for this. We have also suggested solutions where possible.

We produced the SDM process through coproduction. Although 13 people were initially recruited to the coproduction workshops, the number of participants declined with time, possibly due to time constraints and technological issues with online meetings. Participants were not compensated for their time, and we did not seek formal feedback from the drop-in attendees—these are factors to consider for improving retention of participants in the future. However, we believe that the coproduction process was robust, guided by experienced coproduction practitioner researchers and took all underlying principles of this technique into consideration.

Post-pandemic staff shortages prevented us from identifying ward “champions.” The stroke service lacked funding to purchase dedicated iPads, and NHS could not have insured them if purchased with research funds. Ward staff feedback revealed significant time constraints. Implementation may have been possible with SDM champions, ideally senior medics, who could have attended daily ward meetings.

Our study was designed before the COVID-19 pandemic, and the start date was delayed to February 2022. During our study, Treatment Escalation Plans were being implemented in NHS Lothian for all patients and embedded into the electronic medical records. As a result, staff may have felt they were already practicing SDM and saw no need for a new process.

Although Tailored Talks was developed with patient and public involvement in mind, some staff still felt that some materials were not patient-friendly. Additionally, one relative who viewed Tailored Talks reported feeling shocked at hearing the early discussion of prognosis after stroke. Although the main aim of Tailored Talks was to deliver personalized information to patients and their families in various formats to suit individual needs, the feedback highlighted the need to review the materials and remind  staff that patients and families must be ready to receive information after a stroke. Despite general reluctance to use Tailored Talks with patients and families, junior staff found it to be a valuable educational tool for themselves.

We demonstrated the feasibility of using the SURE and CollaboRATE questionnaires in stroke care for the first time, to our knowledge. However, there were insufficient data to determine whether scores differed pre-and post-implementation. Of the 14 patients who died within 6 months, death had been their preferred outcome at baseline. Although we noted an apparent decline in 6-month case fatality for those with severe stroke after the implementation date, it is unclear whether this was related to the SDM process.

Qualitative interviews with family members revealed themes consistent with previous studies, including their experience with stroke and stroke care, diagnosis, discussions about stroke and treatment, involvement in decision-making, and provision of information.4,6,7 Participants provided a wide range of responses, however, a larger study population ideally would have strengthened our qualitative analysis. Although our recruitment efforts were limited by staff and time constraints, even from a small study group, the themes were very similar.

Our experience in developing and implementing a new digital SDM process revealed multiple challenges, which we now aim to address with solutions. Our focus is now on reviewing and disseminating current practice, stressing how SDM is not robustly practiced, engaging senior staff in our stroke unit, and reiterating national priorities for patient-centered care through innovative methods of health care delivery.

Our coproduced SDM process after severe stroke, incorporating the digital communication platform Tailored Talks, was not effectively implemented into practice. Staff shortage was a challenge that led to time constraints and the inability to appoint an SDM champion. Additionally, we encountered technological barriers and recognized the need to re-review  patient-facing information. Given the ongoing workforce limitations and the push to deliver health care digitally, we remain committed to seeking solutions to these challenges and strive to implement an SDM process for patients with severe stroke.

Affiliations:

¹Department of Stroke and Elderly Care Medicine, NHS Lothian, United Kingdom; ²Physical Activity for Health Research Centre, Moray House School of Education and Sport, University of Edinburgh, Edinburgh, United Kingdom; ³Centre for Clinical Brain Sciences, University of Edinburgh, United Kingdom; ⁴University of St Andrews, St Andrews, United Kingdom; ⁵NHS Lothian, Edinburgh, United Kingdom; ⁶Department of Stroke Medicine, NHS Lanarkshire, Glasgow, United Kingdom; ⁷Department of Ageing and Health, Usher Institute, University of Edinburgh, Edinburgh, United Kingdom

Disclosures:

The authors report no relevant financial relationships.

Correspondence:

Akila Visvanathan, PhD

Email: akila.visvanathan@nhslothian.scot.nhs.uk

Acknowledgements

We are grateful to the patients and families who participated, the members of the coproduction group, the clinical staff involved in the management of the patients with severe stroke, and the research nurses Jessica Crossan, Mairi MacDonald, and Sarah Risbridger for recruiting patients. This work was supported by Edinburgh and Lothian Health Foundation Reference 1339.

All authors contributed to the study conception and design. Material preparation, data collection, and analysis were performed by Akila Visvanathan, Gillian Mead, Sarah Morton, Allan MacRaild, Polly Black, and Sophie Gilbert. The first draft of the manuscript was written by Akila Visvanathan, and all authors have commented on versions of the manuscript. All authors have reviewed and approved the final manuscript.