The main work was organised in two Work Packages (WPs) on end of life with dementia, and a third WP on end-of-life research practice.
WP1. Acceptability of the four interventions was studied in a 6-country video vignette study. We developed and pilot-tested four animation videos (see image) to explain the interventions during interviews. Acceptability of euthanasia varied the most between individuals (e.g. 44% of 205 physicians) and countries (up to 66% in the Netherlands, where it is legal). Use of technology was more often acceptable (of the physicians, 66%; up to 74% in the US) and the large majority found ACP acceptable. Qualitative analyses showed that respecting the person’s autonomy was important for physicians, but it could create dilemmas as they also valued appropriate decisions that minimize suffering.
Participants with dementia (n=43) and family caregivers (n=46) showed similar patterns of acceptability. We continue to analyse the qualitative interview data to better understand their perspectives on how the four interventions could be helpful.
WP2. A cluster-randomised trial assessed effects of two approaches of ACP differing in level of control (the scripted medical approach based on shared decision making versus the more flexible coping-based psychosocial approach). General practitioners received two 3-hour training sessions and an e-learning with a conversation with a patient avatar. Participating practices enrolled 21 dyads of persons with dementia and their family caregivers for the medical ACP approach, and 7 for the psychosocial approach; an external control group comprised 54 dyads. Overall changes across groups during 3½ years of follow-up indicated worsening quality of life and improved decisional conflict over time, while families’ self-efficacy to interact, and communicate with physicians did not change. Outcomes did not differ, except for a significant worsening of perceived communication with the physician in the intervention group versus the control group. Possibly there were unmet expectations about continued conversations. Further longitudinal analyses are planned.
Qualitative interviews replaced subgroup analyses to compare approaches and helped refine the theory-based ACP approaches by assessing situations (for whom, when) in which either approach is preferred using data from both WP2 and WP1. Four attributes distinguished the situations: understanding, readiness, trust and momentum (image). For the medical approach, the first 3 were prerequisites for optimal momentum. In contrast, the psychosocial approach would help understand the person, create readiness and foster trust. Without a clear trigger, however, momentum needs to be created. A person-centred approach would entail skill in employing either approach as needed.
Regarding the acceptability of technology, Bispectral Index monitoring was examined in focus groups with 17 persons with dementia and 24 family caregivers. We demonstrated the device. People tried it and didn’t find it bothersome, yet acceptability varied between participants, groups and situations. For example, use in natural sleep was acceptable for most with dementia, not so for most family caregivers. Most would accept use at the end of life if the added value was clearer.
WP3. Ethnographic fieldwork in two end-of-life research groups entailed 12 months of observing how researchers navigate decisions. The ethnographer witnessed researchers’ influence in particular when the research protocol was either under-defined or over-defined, leading to either operationalise or adapt the plans. Research practice involved continued decision making. At times, researchers were unaware of making a decision to adapt plans, potentially limiting transparency. Further, decisions were often rooted in uncertainty due to incomplete information or moral ambiguity.
Greater protocolization cannot solve this. As with ACP, a ‘checkbox approach’ in making decisions in advance fails in educating people on what might happen, and falls short in increasing awareness of their values relevant to future decisions. Rather, researchers should be taught how to navigate decisions. As a team, we put this into practice, regularly discussing research dilemmas.
A proposed Delphi study to synthesize findings was infeasible due to divergent timelines of the WPs, with pandemic-related delays in WP1 and WP2 data collections. We then conducted a literature review and found that, except for Bispectral Index monitoring, technologies to monitor distress had not been tested for use at the end of life. Also, we could conduct an online Delphi study on ACP in dementia with exceptional rigour (image).
We presented findings at conferences and in scientific journals on dementia, palliative, and primary care and research integrity.
