Consumer and Healthcare Professional Led Priority Setting for Quality Use of Medicines in People with Dementia: Gathering Unanswered Research Questions

Establishment of the priority setting partnership

The Priority Setting Partnership (PSP) is the name given to the individuals and groups involved in the process of developing a ‘Top 10’ unanswered question list using the JLA method. Our PSP consisted of a Steering Group, Partner Organizations, a research team, and a JLA advisor. The composition and roles of these groups, and the scope of the PSP is shown in Fig. 1. The Steering Group was composed of 9 individuals (person living with dementia, current carer of a person living with dementia, previous carer of a person living with dementia, geriatrician, general practitioner, pharmacist, registered nurse, nurse practitioner, and social worker). The main roles of the Steering Group for the activities described in this manuscript were overseeing and informing the whole process, deciding on the scope, reviewing project related materials, and contributing to analysis and writing of summary questions. The topic of ‘quality use of medicines in people living with dementia’ was established prior to beginning the study; however, the Steering Group determined the specific scope within this broad area. For example, they decided to include complementary and herbal medicines and education of healthcare professionals. The 9 Partner Organizations (professional and consumer advocacy organizations) supported the project through promotion of the surveys, identification of participants for interviews and the workshop, and will support future dissemination and impact activities.

Fig. 1

Composition and scope of the Quality Use of Medicines in People Living with Dementia Priority Setting Partnership.

Gather evidence uncertainties

The second step in the JLA method aimed to identify as many evidence uncertainties about this topic as possible from stakeholders. Evidence uncertainties were gathered using a cross sectional, anonymous, qualitative survey, administered both online (using REDCap electronic data capture tools hosted at the University of South Australia [34, 35]) and in hard-copy. Participants self-completed (or completed with the help of a friend, family member, or carer) the survey at a time and place of their choosing.

Participants

Two groups of participants were recruited for this study: consumers and healthcare professionals. Consumers included people living with dementia, carers (informal and unpaid carers such as family members or friends), family and friends of people living with dementia (who do not identify as carers), and members of the public with an interest in dementia (such as dementia advocates, previous carers). Healthcare professionals included any healthcare professional (including students and trainees) or person employed by a healthcare organization, with experience providing care for people living with dementia, including those in clinical, support and administrative roles. We targeted healthcare professionals who may have a role directly or indirectly in medicine management, review, or support (see professions listed in Supplementary file 2). All participants were at least 18 years old, able to read and write in English, and Australian citizens or currently living and working in Australia. This work was limited to Australia to ensure that the final Top 10 could be used to influence research, funders, and policy in Australia as many of the potential questions (such as those related to the healthcare system) would be context specific.

Survey design

The survey comprised two sections. Section one asked respondents to list up to three questions and/or concerns about medicine use in people living with dementia that are the most important to them (open question). Asking for three questions was chosen to encourage participants to submit more than one, while minimizing the participants’ anticipated time and burden to maximize completion of the survey (i.e., if the survey asked for up to 10, this may have deterred some from participating). Section two asked optional demographic questions (Table 1). Two different versions of the survey were used for the two different groups, with language and demographic questions tailored towards consumers or healthcare professionals. The surveys were developed based on ones used by previous groups using the JLA method [30], were reviewed by the research team and Steering Group, and piloted with 11 consumers and 5 healthcare professionals with changes made iteratively until they were deemed suitable by the Steering Group (Supplementary file 1: Consumer survey and Supplementary file 2: Healthcare professional survey).

Sample size

As there is no sample size calculation for this type of survey [33], our target sample size of 50–200 responses from each group was based on prior experience of the research team and previous studies that used the JLA method. The decision to stop recruitment was informed by qualitative methodology principles regarding saturation and information power [36, 37] and sufficient responses from a variety of potential participants (informed by participant demographic information) was achieved, while also considering pragmatic constraints such as the study timeline.

Survey promotion

The survey was open from September to November 2020 inclusive. A broad and multifaceted approach to recruitment and convenience sampling was used to obtain responses from a diverse variety of participants across Australia (see Fig. 2). Mostly, the survey was promoted by emailing the advertisement to Partner Organizations and other professional and advocacy organizations, who were asked to distribute the advertisement to their members through the mechanism of their choice (e.g., via regular newsletters, social media posts, at events). Organizations to target were identified by the research team and Steering Group and through internet searches for Australian consumer, professional or other organizations. Additionally, a research engagement service for dementia research known as StepUp for Dementia Research Australia was used to advertise the consumer survey. The study was mostly advertised electronically; however, where requested, paper versions (including reply-paid envelopes) were provided.

Fig. 2

Survey participant recruitment flow chart. Numbers shown are Consumer/Healthcare professionals. Numbers shown here cannot be used to calculate a response rate as it is possible that individuals clicked on the survey to read about it, and then returned to the survey later to complete (which would result in 2 ‘clicks’). Total response rate cannot be calculated as the survey was distributed via external organizations and social media.

Additional data sources

In addition to the anonymous survey, unanswered questions were collected from (i) key informant interviews and (ii) relevant existing national documents identified by the research team (based on their knowledge of this research field). Interview participants were purposively recruited through the same mechanisms as used for the survey but targeted to individuals whose demographic characteristics were underrepresented in survey respondents. Recruitment for interviews was continued until identified gaps in survey respondent characteristics were all filled. The process was iterative, first we reached out to our Partner Organizations and then to other relevant organizations that could identify potential participants. We let them know the types of people we wanted to speak with (e.g., in a specific geographical area, or of a particular profession), the organizations then reached out to people to gain consent for us to contact them. Interviews were conducted by telephone or videoconferencing software, with notes taken by the interviewer (ER or LC, not audio recorded, Supplementary file 3: Key Informant Interview guide) and implied consent used (detailed in the participant information sheet and described verbally by the interviewer before starting the interview). Interviews allowed for in depth discussion and gathering of a large number of evidence uncertainties from a single person to compensate for people with similar characteristics being underrepresented in the survey responses.

Summarizing the responses gathered (analysis)

First, the responses to the survey were reviewed (by ER) and those that were out of scope (such as not medicine-related or related to the development of new medicines or non-drug care, Fig. 1), or not considered to be potential research questions (such as unclear submissions, or questions asking for information or advice) were removed. Second, conventional content analysis was used to analyze and group (categorize) the data [38]. In this approach, related sections of text were compared and put into groups based on conceptual relationships (i.e., those that were reporting the same concepts). Third, each group of related responses were reviewed, and an indicative summary question was generated. Fourth, summary questions were grouped into themes. This was then repeated with the interview notes and key documents, in a single analysis process (i.e., one set of results).

The conventional content analysis was conducted by a single investigator (ER), with the themes and summary questions reviewed by, and discussed with two members of the research team (MS and TN) on two occasions during the process of data collection. At the end of the data collection process, the Steering Group reviewed the themes, summary questions, and quotes which informed the questions; changes were made iteratively until consensus was achieved by the Steering Group (via virtual meetings and email). This could involve creating, altering, merging, or removing groups (i.e., summary questions) in alignment with the conventional content analysis process. The Steering Group also reviewed the quotes that were removed due to being out of scope or not research questions (deciding if any needed to be added back into the conventional content analysis) and could suggest new questions to be added based on their own lived experience that had not arisen from the data sources. QSR Nvivo 12 Plus software was used for data management, coding, and facilitation of the analysis. Participant characteristics were analyzed descriptively to describe the participant population.

Evidence checking

Following analysis, the next step involved determining if the summary questions were ‘unanswered’, that is, whether a level of uncertainty still exists about each question. Uncertainty was defined as no up-to-date, reliable systematic reviews or evidence-based guidelines of research evidence, or alternatively, up-to-date systematic reviews/guidelines reported that uncertainty existed. At this point (prior to evidence checking), no further questions could be added, and questions were removed or altered as a result of evidence checking.

Each summary question was checked by a member of the research team (ER, LC, MS, LKE, JG-T, ET, JKS, KT, SNH; referred to as ‘information specialists’ in the JLA method) using relevant key words in the following databases: Cochrane Database of Systematic Reviews, National Institute for health and Care Excellence (NICE) Guidelines, and Australian National Health and Medical Research Council (NHMRC) Guidelines Registry. At the discretion of the information specialist, professional organization websites, PubMed, and Epistemonikos were also searched where relevant. References from 2018 until date of search (April 2021) were included. If the information specialist was unsure about whether a question was unanswered, this was resolved by discussion with another member of the research team and decided by the Project Lead when required. The research team had the opportunity to review all changes and removal of questions; however, no formal consensus process was conducted. The full details of this process are available online [39]. This rapid review approach aligns with the established JLA method, which recommends a pragmatic and proportionate evidence search [33]. Where it was identified that the summary question was ‘answered’ it was then removed from the list, to end up with a list of ‘unanswered questions’. Additionally, where the question was determined to be partially answered, it was reworded to focus on the unanswered element, with review of responses that informed the question to make sure that the summary question still represented the original data.

Participant characteristics and additional data sources

A total of 151 consumer participants and 77 healthcare professional participants completed the survey and were included in the analysis (Table 1, Fig. 2). Most consumers identified as a family member or friend (42.5%) or a carer (26.0%) of a person living with dementia with about 10% identifying as living with dementia. The most common healthcare professionals who responded were registered or enrolled nurses (34.2%) and pharmacists (14.5%). Most participants were female (consumers: 77.0% and healthcare professionals: 71.1%) and born in Australia (74.0% and 69.6%, respectively).

Eight key informant interviews were conducted with individuals with characteristics that were underrepresented in the survey; specifically, people living in three Australian states/territories (Tasmania, Queensland, Northern Territory); geriatricians; participants who had significant experience working in residential aged care facilities or worked in a practice that solely focused on caring for people living with dementia; Aboriginal and Torres Strait Islander Healthcare Workers and Practitioners; people who identified as Aboriginal and Torres Strait Islander; and those with experience in rural and remote communities (with most key informants fulfilling more than one of these characteristics).

Finally, an additional four existing documents were identified by the research team as relevant: a clinical practice guideline for people living with dementia, and recent reports from Dementia Australia, Alzheimer’s Disease International, and the Australian Royal Commission into Aged Care Quality and Safety [40–43].

Summary questions and themes

The 228 survey participants submitted a total of 545 questions (each participant was encouraged to list up to three). Analysis of the survey responses, key informant interviews and existing relevant documents were synthesized into 67 questions, and the Steering Group added one additional question during their review. These 68 questions were grouped into 13 themes (range of 1–12 questions in a theme, Table 2). No additional summary questions were created after the sixth key informant interview, or analysis of the four relevant existing documents (i.e., data saturation was achieved). Of the 68 questions identified, 59 questions were informed by both consumers and healthcare participants, six were informed by consumers only, and three by healthcare participants only. Summary questions were informed by between 1 and 83 unique respondents/sources. One question was solely informed by a single key informant interview.

The themes (number of questions) were: Awareness and Education (3 questions), Changed Behaviours (3), Healthcare System and Person-Centred Care (7), Residential Aged Care Facilities (2), Medicine Management (6), Polypharmacy, Complexity, and Deprescribing (8), Treatment of Dementia (9), Specific Comorbidities (11), Specific Adverse Drug Reactions and Harms (12), Monitoring for Harms and Benefits (1), Pharmacokinetic and Pharmacodynamic Changes (1), Influences on Prescribing (2), and COVID-19, Vaccines and Restrictions (3).

Evidence checking

The 68 summary questions underwent evidence checking (see Supplementary file 4: Evidence checking results). No questions were judged as ‘answered’; however, four questions were determined to be partially answered and so were revised to focus on the unanswered aspect of the question (Supplementary file 5: Questions altered after the evidence checking process). One question (Question 23: ‘What medicines are commonly used in people living with dementia?’) was removed during the evidence checking process, and this deletion was agreed by the Steering Group, as it did not focus on a specific issue that could have a direct impact on clinical practice.

This resulted in a final list of 67 unanswered research questions about quality use of medicines in people living with dementia, as suggested by consumers and healthcare professionals in Australia.

Summary of findings and comparison with previous research

This study represents the first application of the JLA method in an Australian context and for a research topic focused on ‘quality use of medicines’. The research identified 67 unanswered research questions, grouped into 13 themes, gathered from people living with dementia, their informal carers, family members and friends and the healthcare professionals that care for them. This work highlights the importance of asking people directly impacted by dementia and those caring for them about uncertainties with regards to medicines use and what research they want to inform optimizing the health and well-being of people living with dementia.

The grouping of summary questions into themes was performed primarily to assist with the review and presentation of the questions. However, the wide breadth of themes arising from the data shows the variety of different research areas within the field of quality use of medicines in people living with dementia. We identified themes about the awareness and education of both consumers and healthcare professionals, the healthcare system, the setting of residential aged care facilities, management of medicines in the home, polypharmacy, treatment of dementia, changed behaviors, and other medical conditions, specific adverse drug reactions and harms and others.

None of the summary questions identified in the present study were determined to be fully ‘answered’ in the literature. This was unsurprising as quality use of medicines in people living with dementia is an under researched and complex area [44]—both in the relationship between medicines and dementia and appropriate and safe use of medicines in this population. Clinical drug trials tend to exclude people with dementia, those who have more complex health issues (e.g., multiple medical conditions) and/or in later stages of the medical condition [45, 46]. It cannot, therefore, be assumed that the results of clinical trials are generalizable to people living with dementia, particularly those taking multiple medicines [47]. Evidence suggests that this population is at a higher risk of medicine-related toxicity due to changed pharmacokinetics and pharmacodynamics, but the clinical importance of this is mostly unknown [48, 49]. Similarly, people living with dementia are often excluded from health services research owing to perceptions of inability to consent to participate in research, or take part in the processes required for the study [50, 51]. For example, Taylor et al. [52] found that 16% of original research in a leading geriatrics journal used recruitment methods that would prevent participation of people with cognitive impairment, and 29% of publications explicitly excluded this population. The authors also noted poor reporting regarding whether and how people with cognitive impairment were excluded [52].

Two previous groups (in the UK and Canada) have used the JLA method to conduct priority setting work on the broad topic of dementia. In their work, a small number of medicine specific questions arose, such as how to best treat (pharmacological and non-pharmacological) changed behaviors, long-term effectiveness of cholinesterase inhibitors, and how to support independence (including with respect to medicine-taking). While the majority of their unanswered questions did not explicitly mention medicines, many similarities exist with our themes and unanswered questions, such as aiming to achieve person-centered care, optimizing quality of life, staff training, and supporting people living with dementia to remain at home [31, 32].

It has been established that there is an important mismatch between research being conducted and research that consumers and healthcare professionals think should be prioritized. Crowe et al. compared the ‘Top 10’ priorities of the first 14 JLA Priority Setting Partnerships to a random sample of registered clinical trials over the same period. Medicine treatment was the focus of 86% of commercial registered trials, and 37% of non-commercial trials, yet was only the focus of 18% of consumer and healthcare professional priority research questions [3]. While our Priority Setting Partnership focused on the field of quality use of medicines, we similarly found that there were some questions related to the pharmacological treatment of dementia, but the majority covered a broad range of topics including the management of medicines, training of healthcare professionals, and how to optimize medicine use and minimize medicine-related harm.

Strengths and limitations

This study used a robust methodology developed by the JLA and employed for over 100 different health topics internationally. Additionally, the flexibility in data collection for the first survey enabled us to maximize responses from consumers and healthcare professionals across Australia.

There are several limitations to this study. First, we were limited to mostly distributing the survey and advertising the study online and our convenience sampling approach may have limited capturing the voices of people who are unfamiliar or unable to complete an online survey in English (such as people with moderate to severe cognitive impairment and people from culturally and linguistically diverse backgrounds). As this study was conducted during the COVID-19 pandemic we adopted a risk minimization approach and chose not to conduct recruitment or data collection in person (considering the increased risk for people living with dementia [17]). We identified some groups as underrepresented in our online survey, such as those in a non-clinical role (e.g., administrators of residential aged care facilities) and people living in the Northern Territory. To address these gaps, key informant interviews were used, and existing documents reviewed. However, use of primarily Partner Organizations to recruit the key informants may have led to bias towards inclusion of people who are more informed about quality use of medicines issues. Some of the demographics of our population do reflect the broader target population, such as one third of the Australian population being culturally and linguistically diverse (26/30% consumers/healthcare professionals respectively of our respondents were not born in Australia), and the majority female respondents reflecting that women make up the greater part of the health and social sector workforce (especially pharmacists and nurses) [53], and that women make up the majority of carers [42]. Second, inherent limitations related to the survey methodology included not being able to determine a response rate, having no method to limit duplicate participation by a single participant, and not adjusting for missing data or non-responses. However, as our analysis was qualitative, and driven by the aim of identifying the broad range of possible research questions from a variety of people, these limitations are only likely to impact data concerning participant demographics. Third, we did not identify any new summary questions after the sixth key informant interview or after review of the relevant existing documents which could be considered as reaching data saturation. However, there were several summary questions which were only informed by a single respondent (and therefore may not fulfill other definitions of saturation such as degree of development of a code) [37]. Finally, while we employed a rapid review approach to evidence checking which is not as robust as a systematic review process, it was appropriate for this study given the high threshold for a question to be answered (i.e., if there was any doubt that the question had been answered, it was retained and considered an unanswered research question). Indeed, no questions were removed due to being answered. Additionally, the summary questions are not necessarily a single research question that could be answered in a single study and therefore, researchers acting on these unanswered summary questions will still need to review the existing literature when planning their future studies.

Research, clinical, and policy implications

National and international initiatives have highlighted the importance of ensuring quality use of medicines in older adults and people living with dementia. In June 2019, the World Health Organization (WHO) released their report of ‘Medication Safety in Polypharmacy’ as a part of the current global patient safety challenge of ‘Medication Without Harm’. The WHO highlight that people living with dementia are a population to be targeted for medicine reviews as they are at risk due to polypharmacy [13]. In Australia, a Royal Commission into Aged Care Quality and Safety was established in October 2018 in response to concerns about the care of people living in residential aged care facilities, including the high prevalence of inappropriate antipsychotic medicine use in people living with dementia [43].

Solving national and international issues of medicine overuse and medicine-related harm will require research grounded in practice and informed by those who are most affected by them. The unanswered research questions identified in this study highlight areas of concern (such as medicine-related harm and the overuse of antipsychotics) and seek to find answers on how to solve these problems. For example, the theme of Changed Behaviours includes three questions that could direct the path of future research, including: asking when and how these medicines should be used (considering the type and stage of dementia); the most effective and safe way to reduce the use of antipsychotics; and if certain medicines increase the risk of changed behaviors. Additionally, questions in the Healthcare System and Person-Centred Care theme (such as how to achieve informed consent and shared decision making), Residential Aged Care Facilities theme (such as how to provide aged care staff with necessary knowledge and skills) and Polypharmacy, Complexity and Deprescribing theme (such as preventing prescribing cascades, optimal model of medicine reviews and how to achieve deprescribing) provide insight into how these issues could be addressed.

Conclusion

The findings from this study have great potential to provide benefits to people living with dementia and their carers by guiding researchers to align future studies in the field of quality use of medicines in dementia with the highlighted unanswered consumer and healthcare professional questions. Moreover, findings provide a consumer- and healthcare professional-centered foundation for researchers who are designing future research, while not replacing the need to include consumers and healthcare professionals in individual projects. By creating a list of unanswered questions relating to quality use of medicines driven by key stakeholders, future research will be inherently linked to optimizing care and quality of life for people living with dementia. Future work will involve further engagement via second survey and national workshop with consumers and healthcare professionals in Australia to determine the top 10 unanswered research questions about quality use of medicines in people living with dementia.