A co-design process developing heuristics for practitioners providing end of life care for people with dementia

Using information from three sources and applying an iterative co-design approach, triangulation of data was carried out using a literature review, qualitative interviews and focus groups with family carers (‘experts by experience’), and focus groups with health and care professionals. Interviews were only conducted with family carers as some family carers expressed that they did not wish to take part in focus groups. Co-design is a technique adopted from product development [36] which has tangible benefits in developing or redesigning health services [37‐40]. The co-design task was to develop heuristics with only few attributes that operate under three rules: 1) search through cues in a pre-determined order; 2) stop searching as soon as a cue leads to an exit; and 3) classify the object of concern accordingly [35]. An example is shown in Fig. 2, showing a decision tree for treatment of community acquired pneumonia attributable to Mycoplasma pneumoniae in children with macrolide antibiotics [28]. The branches of the tree show the pre-determined order of questioning, the object of concern is Mycoplasma pneumonia and three levels of risk represent the exits from the decision tree.

An overview of the procedure is shown in Fig. 3. A rapid appraisal literature review was conducted to identify key areas of decision-making concerning people with dementia at the end of life [42]. The findings of the review, together with findings from a preliminary study which interviewed family carers about their ideas of quality end of life care for people with dementia, [18] were used to develop a topic guide (see Additional file 1).

The topic guide was used to facilitate focus groups and interviews with both former and current family carers of people with dementia, as well as groups of health and care professionals. The topic guide for the interviews and focus groups explored six broad areas of difficult decision making; 1) difficulties with swallowing and problems with eating, 2) agitation/comfort, 3) ending life sustaining medical treatment, 4) personhood, 5) stopping routine care (i.e. bathing and turning of a bed bound person), and 6) communication between professionals. Additionally, carers were asked to suggest further difficult decisions which needed to be made at the end of life.

The focus groups used a ‘think aloud’ strategy; participants were presented with each topic and the group were asked to vocalise their thoughts about how decisions regarding this topic are currently and should be made [43]. Traditional semi-structured interviews give people the time and chance to process and rationalise thoughts and decisions, and therefore do not necessarily give an indication of what they really think or provide any understanding of how they arrived at their decision [44]. In order to make sense of what people think and why they make the decisions they do, many have argued that we need to pay attention to verbalizations and thought processes as these highlight how information is stored in working memory [43, 45, 46]. Working memory contains the thoughts in the present moment as well as the memories which are being retrieved from long-term memory [47]. The think aloud strategy allows us to hear and access thoughts in working memory and hence get a better understanding of how individuals make decisions. The ‘think-aloud’ method was developed for the usability testing of technical products but has been used to understand the complex process of decision making in healthcare settings [45].

Findings from the interviews, focus groups and the literature review informed the development of a toolkit of heuristics. Transcripts were read by four members of the research team (ND, SI, RM, JW) and key messages and decisions/decision processes were recorded by each researcher. The research team met to discuss all key messages from the data and how these may reflect heuristics. From this initial meeting a series of heuristics consisting of statements were generated. Two members of the team (ND and RM) constructed heuristics from these statements as flowcharts as shown in Figs. 4, 5, 6 and 7. The research team met again to discuss the developed heuristics in order to refine them, using the interviews and review of the literature as a basis for this, together with their clinical and research experiences.

To develop the heuristics further a co-design group was established. The co-design group consisted of health and social care practitioners (palliative care consultant, two GPs, Admiral (specialist dementia) nurse, social care professional, two geriatricians, and a community nurse), four family carers and members of the research team with backgrounds in psychology, social care, anthropology, and general practice. The co-design groups’ expert knowledge and interpretation of the available evidence was used to further develop the heuristics prior to their testing in real settings.

The co-design group met following analysis of findings from the focus groups and interviews to discuss the development of the heuristics, using a nominal group process. Nominal groups are structured meetings with an aim to facilitate group thinking and decision making about a given problem within an expert group. A nominal group process was chosen as it has been shown to have a useful role in analysing health care problems [48], and can help bridge the gap between researchers and practitioners [49]. Such groups keep experts’ time commitment to a minimum and collect a variety of ideas whilst allowing for interaction and discussion of these ideas.

The group were presented with an overview of the evidence from both the literature review and qualitative data, using a PowerPoint presentation. Each member of the group was given an opportunity to ask questions of clarification and detail from the team. Following the presentation each member of the team was provided with the provisional heuristics, as both written statements and flowcharts. Members of the group were initially asked to read through these independently and then bring thoughts back to the group about the presentation of these heuristics. Specifically members of the group were asked their preferred format: visual image or written statements. Following this, each topic was discussed in detail. The research team presented each topic individually and asked specific questions of the group to promote discussion and facilitate problem solving. Examples of questions and the answers to these questions, with implications for changes to the heuristics, are discussed below.

Detailed notes were taken of the nominal group process with the co-design group of comments and discussions which were made and the conclusions drawn from these discussions. These notes were then used by the research team to inform a rapid iteration following the meeting. The nominal group outputs were sent by email to all members of the co-design group for further comment and feedback on the heuristics. A near-final prototype of each heuristic was revised by the research team and sent to the members of the co-design group, prior to testing in real settings.

Key topics regarding difficult decision making were considered to be: eating and swallowing difficulties, agitation/restlessness, ending life sustaining treatment, and providing routine care at the end of life. Other topics, including person-centred care and communication between professionals, were deemed by the co-design group to be less appropriately addressed by heuristics and also more ‘generic’ to good dementia care rather than specific to end of life care.

Heuristic three consists of two rules which may help with decisions for ending life sustaining treatment and initiating new forms of treatment (see Fig. 6). Life sustaining treatment was considered to be any treatment that was not solely about symptoms and that had long-term benefits, such as statins or antihypertensives. The first rule encourages an iterative process of discussions with family members and a multi-disciplinary team approach, whereby treatment is only continued or initiated if it maintains comfort or quality of life.

The second rule focuses on how to proceed with rationalising medication and determining what treatments are still needed at the end of life. Decision-makers are advised to only continue or initiate treatments that are likely to maintain comfort or have an impact on quality of life. The rule advocates removing treatments one by one, in order to avoid ambiguity about any possible adverse effects following their withdrawal. Subsequently, following the removal of any treatment, quality of life and comfort should be reassessed and the professional should be prepared to reinstate treatments if appropriate.

Discussions with the co-design group did not lead to significant changes in this heuristic. There was a group emphasis that this heuristic should not include antibiotic treatment. It was felt that whether or not to initiate antibiotics was an acute treatment decision that required its own decision-making framework. Specifically, it was not felt to be comparable with decisions about whether or not to continue statins and antihypertensives.

Providing “routine care” encompassed several aspects including personal care (i.e. bathing and grooming) as well as turning an individual who is bed bound as a preventative measure for pressure ulcers (see Fig. 7). However, routine care does not include turning or moving a patient if they already have a pressure ulcer and need to be turned to relieve discomfort. This rule, unlike many of the others, applies to the final few days and hours of life, when providing routine care may be uncomfortable and unnecessary for the individual.

The heuristics consists of three rules which prompt professionals to discuss with the family or advocate about what an acceptable level of care may be. For example, there should be discussions with relatives about whether the patient would like their bed sheets to be changed if they were not wet or soiled (wet and soiled sheets/clothing would always still be changed).

The second rule which will be informed by these initial discussions encourages the professional to consider if routine care is causing distress and if so make amendments or provide an agreed minimal level of care without causing distress. Finally, rule three highlights that if distress due to routine care is recurring, then further discussions with the family about an acceptable level of care are recommended.

The co-design group felt it was important to emphasise at the start of this heuristic the need to discuss with the family/advocate about what they feel is an acceptable level of care. The original heuristic presented to the co-design group placed discussions with the family at the end after problems around providing routine care were identified. The co-design group also felt this heuristic was better suited to being deconstructed into three separate rules.

We used an iterative co-design process to develop a toolkit of four heuristics on important and complicated problems which practitioners working in end of life care for people with dementia find challenging. We combined information from the existing literature, together with the perspectives of family carers and practitioners from a range of disciplines, backgrounds and experience in a technology development method to synthesise evidence based heuristics.

The four heuristics developed in this study have been developed with the aim of being used by health and care practitioners, including doctors, nurses and care practitioners in different settings. They provide a flowchart of rules (accompanied by additional written material) to help practitioners make informed decisions with more confidence.

We would suggest that the heuristics have the potential to be used alongside the international EAPC recommendations of optimal palliative care for people with dementia [17]. The EAPC recommendations provide a comprehensive overview of care and what palliative care for someone with dementia should look like, whilst heuristics provide more practical assistance to guide practitioners through making difficult and complex decisions.

The development process raised questions about which audiences and settings are the most appropriate for the use of heuristics. As others argue, some heuristics will be more appropriate for nursing and social care and others for medical care [51]. The same applies to the setting in which these heuristics may be appropriate, for example providing routine care would be applicable across all relevant settings, community, care home, hospital or hospice. However, aspects of the eating and swallowing difficulties heuristic may not be applicable to community teams, who are unlikely to insert nasogastric (NG) tubes in the home, although they may support community dwelling people with them in situ, although this may also vary across different countries. Research from the USA suggests there is no evidence that artificial feeding increases quality of life, [52] and a general consensus from International EAPC recommendations that permanent artificial feeding should be avoided [17].

The discussions about heuristics in this development process have also highlighted the need for caution in their use. Although heuristics have been described as a form of ‘fast and frugal’ decision making which frequently leads to the right answer [53], not all end of life decisions can be made quickly. Some decisions will require careful consideration, good communication within teams and between professionals and families, checking advance care plans and a multi-disciplinary team approach - for example, stopping life sustaining treatment. However, heuristics can in some instances help create a logical framework for making fast decisions which lead to action, such as the provision of routine care.

Some of the heuristics such as agitation/restlessness and eating/swallowing difficulties assume that specialist help such as a specialist palliative care team, geriatrician, or old age psychiatrist is available, however not all teams particularly in the community where many people with dementia reside will have access to such help. For example, the heuristic regarding agitation/restlessness suggests seeking specialist help from an old age psychiatrist if there is no identifiable cause of agitation/restlessness. However, again this may vary across countries and health care systems.

Despite the attraction of heuristics being a simple, efficient, fast method particularly useful for conditions of uncertainty, [28] they are mental shortcuts and therefore there are limitations in their use; in particular they can lead to faulty conclusions and cognitive biases [54]. One particular example of cognitive bias is the concept of ‘representativeness’, which refers to errors made because thinking is overly influenced by what is typically true. For instance, when someone with advanced dementia begins to refuse food, this may be due to the disease process itself but it can also be due to an entirely separate physical or even psychological cause, which may be overlooked and reversible, for example oral thrush. The heuristics presented in this paper are transparent and therefore minimise error in judgement and resulting bias, but they do not eliminate error. Throughout the heuristics process, the importance of discussions with the family, nominated decision maker and person with dementia (if able) is highlighted. Discussions should happen as early as possible as part of an approach to advance care planning. However, this may not always be possible as some people with dementia and their families do not wish to discuss their own mortality and end of life care plans [18, 19, 55].

A strength of this study is the direct focus it has placed on the views of family carers and those who make difficult decisions about end-of-life care for people with dementia. The co-design group contained a mix of health and care practitioners and former family carers which allowed for detailed discussions about the heuristics and their content, increasing the rigour in the development process which has been highlighted as often lacking in previous research within this field [26].

This study is limited by the number of rounds (two) of iteration that the heuristics received; more rounds of discussion with the co-design group may have enhanced the usability and design of the heuristics. However, further development work is planned in subsequent phases of the study.

The developed heuristics are being used by five different teams including one general practice, one community nursing team, an older adult hospital ward, and two community palliative care teams [20]. The heuristics will be used in practice by these five teams and evaluated after three and 6 months of use. Some alterations and refinements of the heuristics will likely be made with the co-design group using a similar procedure to that described in this paper at the 3 month stage. Finally, at the close of the study, the co-design group will help produce a finalised version of the heuristics.