Advance care planning in dementia: a qualitative study of Australian general practitioners

Keywords: advance care plan, advance care directive, dementia, cognitive impairment, general practice, qualitative research, theoretical domains framework, implementation science, patient-centred care, decision-making capacity.

Advance care planning (ACP) is a process of reflection, discussion and communication that enables a person to plan for their future medical treatment when they will no longer be able to make decisions (Royal Australian College of General Practitioners (RACGP) 2012). It leads to increased concordance between preferred and delivered care (Houben et al. 2014). ACP is particularly relevant for people with dementia because the condition eventually precludes the affected individual from taking part in their own medical treatment decisions.

GPs play a key role in dementia ACP (Bally et al. 2020), yet only 3.2% of audited records of patients aged ≥65 years in Australian general practice had a formal advance care directive (ACD) (Detering et al. 2019). Reasons behind the low uptake of ACP within the context of Australian general practice remain understudied. A recently published systematic review on the barriers and facilitators for GPs in dementia ACP did not identify any study of Australian GPs (Tilburgs et al. 2018). The aim of this study was to address this gap by exploring factors that influence ACP among Australian GPs caring for people with dementia.

A qualitative study design informed by the theoretical domains framework (TDF) (Atkins et al. 2017) was used. The TDF was developed through an international collaboration of behavioural scientists and implementation researchers. In this study, the TDF provided a framework to comprehensively investigate aspects of dementia ACP that may be amenable to intervention. We followed published guidance on how to operationalise the TDF and developed an interview schedule drawn from the 14 domains of TDF version 2 (Atkins et al. 2017) (Table 1). Subject-specific questions for each domain were initially adapted from existing examples in the literature (Michie et al. 2005; Huijg et al. 2014) and further refined after conducting two pilot interviews with GP researchers. Data from these pilot interviews were not retained for analysis.

Table 1.  Example questions from the interview guide based on the TDF

The study was led by an academic GP registrar (AA). An experienced qualitative researcher (CB) provided advice and close guidance throughout the research process. A recently fellowed GP with experience conducting qualitative research and undertaking a PhD (PP) helped with data analysis and write up. The final author (DM) was the lead supervisor responsible for steering the overall direction of the study.

Initially, a random sample of 500 GPs registered with the Monash Practice-Based Research Network (MonReN) were approached to participate in the study. The MonReN database comprises 1800 GPs working across the south-east region of metropolitan Melbourne, and the list was randomised using Microsoft Excel. Each GP received a postal invitation to participate that included the explanatory statement and consent form. Prospective participants had the option of returning the signed consent form via an enclosed reply-paid envelope, email or fax. A reminder was sent approximately 6 weeks after the initial mail-out.

We received 30 signed consent forms, a response rate of 6.6% after excluding 47 undeliverable letters (Fig. 1). Respondents were contacted to arrange one-on-one semi-structured interviews, which were scheduled in the order that the consent forms were received, up to the 13th interview. The remaining participants were purposively sampled for maximum variation of sexes. The stopping criterion was data saturation (Francis et al. 2010), which was reached by the 16th interview.

Fig. 1.  Flow chart for recruitment.

Interviews were conducted by one of the authors (AA) over the telephone for participant convenience. We used in-depth interviewing principles (Minichiello et al. 2008) with open-ended questions to encourage participants to share their unique personal narratives and experiences. In addition, field notes were taken (by AA) after every interview and a reflective journal was kept. After initially undertaking three interviews, transcripts and notes were reviewed by two authors (AA and CB) to ensure the interviews were working well. Data from these three interviews were included in the final analysis.

Interviews were conducted between August and November 2019, digitally recorded and transcribed verbatim by an external transcription service. Unedited transcripts were mailed to respective participants along with an A$200 gift voucher. As part of member checking, all participants were given the opportunity to review their transcript for accuracy, to make clarifications or expand on what they had described in the interview. Two participants returned their transcripts with changes, which were minor clarifications.

Data were analysed in two steps. First, data were analysed deductively to determine the most pertinent domains and then inductively to identify overall themes (Atkins et al. 2017). Two authors (AA and PP) independently reviewed the first five transcripts and participants’ responses were distributed among the previously determined theoretical domains within the framework, through directed content analysis. Transcripts were read line by line, with whole segments of text assigned to relevant domains. After coding text from the first five transcripts to one or more of the 14 TDF domains, discrepancies were discussed and resolved with input from a third author (CB) to increase rigour. A final coding guideline was developed by the team, and the remaining transcripts were then analysed by a single author (AA).

Once text from each transcript had been assigned to domains, the contents were further analysed by a single author (AA) following the steps of thematic analysis (Braun and Clarke 2006). Phase 1 of gaining familiarity with the data occurred through the process of conducting the interview, referring to the field notes and deductive analysis of the transcripts. Instead of generating novel codes, domains of the TDF served as the initial coding framework for Phase 2. The remaining Phases 3–6 of the thematic analysis methods of Braun and Clarke (2006) were followed as per published guidance. All data were managed using NVivo 12 Plus.

The study was approved by the Monash University Human Research Ethics Committee (Project ID 18800). Participants provided written consent before interviews and confidentiality was assured by deidentification. The external transcription service also adhered to a confidentiality agreement.

Sixteen GPs were interviewed (eight males, eight females). The duration of the interviews ranged from 32 to 70 min, with a mean of 44 min. One participant’s primary place of work was an Aboriginal Medical Service. Years spent in general practice ranged from 1 to 25 years (Table 2). Data on age, ethnicity and country of graduation were not collected because time in practice was deemed to be more relevant to interpreting the data.

Table 2.  Participant characteristics

The most prevalent domains were: (1) environmental context and resources; (2) beliefs about consequences; and (3) social/professional role and identity. This prevalence was determined by the number of times interview content was ascribed to a domain within the TDF. Table 3 contains illustrative quotes for each of the 14 domains and Table 4 maps the major themes to relevant TDF domains. Three major themes were identified from the inductive analysis, namely the role of the GP, the contribution of ACP in promoting patient-centred care and the burden of responsibility, and these are described in detail below.

Table 3.  Domains and themes
CALD, culturally and linguistically diverse; MBS, Medicare Benefit Schedule

Table 4.  Summary of themes and links to the TDF

GPs felt that engaging patients in ACP early was a big part of their role, especially before the onset of dementia. In addition, practice nurses may assist in bringing ACP to the attention of patients as part of routine health assessments. Furthering the conversation and following through with the paperwork was still seen as the responsibility of the doctor:

Well I think it’s massive, it’s massive…I think putting into place plans before people get dementia, while they’re of sound mind, probably in their aging is a good idea. And of course when people do have dementia, putting in plans and having plans in place is a very good idea too. So I think pre-planning, and planning at the time. GPs are critical because especially in the pre-planning stage we know the patients generally really well [GP16, male, 22 years post-fellowship].

Our participants felt they possessed sufficient knowledge and skills for ACP. They understood ACP as an iterative process and identified rapport, active listening, clear communication and empathy as key to success with ACP. However, depending on the stage of dementia, discomfort in assessing a patient’s capacity posed a major challenge to ACP and often led to specialist involvement:

I think I’m moderately confident about advance care planning, but I’m incredibly unconfident [sic] about advance care planning in patients with dementia. I think that’s a very difficult situation – difficult enough, I think that I probably would secede in many situations to a geriatrician [GP4, male, 1 year post-fellowship].

Participants acknowledged the role of other healthcare providers. Hospital-based junior doctors, nurses and social workers may contribute to ACP in primary care by triggering a conversation during an inpatient encounter that continues with the GP after discharge from hospital. The challenge of navigating ACP in a culturally appropriate manner means that cultural and Aboriginal health workers are likely to have increasingly important roles in some settings.

The belief that ACP would help align the goals of patient, carers and clinicians emerged as a significant motivator for GPs. As one participant put it:

After her mum passed away she said to me having that advanced care planning form was the best thing because they were able to go in when the mother got admitted, show the admitting doctor this is all the paperwork we have written. We clearly don’t want anyone to do this or that. We just want to be kept comfortable and so even though her mother doesn’t speak English she didn’t worry [GP12, female, 18 years post-fellowship].

Although a strong conviction in the usefulness of ACP was shared among most participants, there was some pessimism about the utility of ACP documents, such as formal ACDs. Some GPs thought hospital clinicians may not be able to gain access to an ACD to guide critical decision making or an ACD may lack sufficient detail to assist with all clinical scenarios. There were also concerns regarding possible damage to the doctor–patient relationship if approached insensitively:

The main risk for me is the relationships with the family and patient. If you don’t do it sensitively enough you will lose them – you will lose their trust. The other thing I think is a risk is if you don’t do it well the patient doesn’t get what they want in the hospital or where they are [GP6, male, 19 years post-fellowship].

GPs commented that within the standard consultation ACP was most likely to proceed if initiated by patients, but few patients present specifically with the goal of ACP in mind. Perceived patient reluctance was often a barrier:

I have brought up the issue with patients…there’s often a bit of awestruckness to it, they’re sort of hit from left field. They don’t feel that it may be appropriate for them at the time, so you’ve got to come back to them in a year or whenever [GP7, male, 17 years post fellowship].

GPs expressed a desire for patients to take greater responsibility for initiating ACP. They believed that improved patient awareness on the benefits of ACP would encourage more patients to visit their GP for ACP. To this end, several participants recommended increased public health messaging to overcome societal taboo around discussions that require individuals to grapple with the concept of death and dying. There was also a sense that increased availability of culturally and linguistically diverse patient education resources may also assist with raising awareness about ACP in today’s multicultural society.

Our practitioners experienced greatest success with ACP when it was incorporated into existing preventive care workflow, such as health assessments. Some GPs were also concerned about the out-of-pocket cost to patients who may need multiple consultations for ACP. Even when lack of funding was identified as a problem, participants were ambivalent in their support for a new Medicare item number dedicated to ACP; although some believed it would incentivise more GPs to initiate ACP, others were worried it would dilute the quality of ACP:

I’m sure if the [Medicare Benefit Schedule] introduced an advance care planning item number, suddenly every brother and his mother would have an advance care plan. I don’t think that’s actually something that I would advocate for myself, because I feel that were the intentions clouded or muddied that way, you would have advance care plans…not worth the paper they’re written on [GP4, male, 1 year post-fellowship].

In addition to the patient factors and systems-level barriers mentioned above, some participants admitted that ACP was not necessarily on their agenda. A range of competing priorities often draws the GP’s attention away from ACP unless there are clinical triggers, like progression of chronic illness or a significant new diagnosis. Participants mentioned a role for external reminders, such as articles about ACP appearing in journals and talks from those interested in increasing the uptake of ACP:

…it’s not actually brought to our attention…if you want me to pay attention, then send me the propaganda, because that’s what makes me go, oh, yeah, that’s a good idea [GP14, male, 12 years post-fellowship].

This study found that Australian GPs see themselves as having an important role in facilitating ACP, especially before the onset of dementia. It confirms previously reported barriers of patient reluctance to engage in ACP and clinician lack of time (Rhee et al. 2012; Scott et al. 2013; Howard et al. 2018; Tran et al. 2018), but also highlights that GPs’ beliefs about the benefits of ACP can act as an enabler.

To our knowledge this is the first study to explore ACP in Australian general practice using the TDF, and our use of an implementation sciences framework may assist in developing interventions to increase the uptake of ACP in dementia in primary care. Of the 14 TDF domains, the ones we found to be most relevant to the practice of ACP correlated with those of a scoping review by a research group in the US (Nelson-Brantley et al. 2020), which also found social/professional role and identity, as well as environmental context and resources, to be important determinants in initiating ACP.

This study provides valuable insights from a sample of GPs motivated to improve ACP in primary care. Among our participants there was a strong sense that patient factors were a key barrier to ACP. GPs advocated for public health campaigns to raise patient awareness of the benefits of ACP, a perspective supported by the literature (Rhee et al. 2012; Scott et al. 2013). Efforts in this domain should be culturally specific so that they can permeate various minority groups. Although lack of time and funding were also identified as barriers at the systems level, designing interventions to address this perennial issue in primary care is challenging.

The debate around benefits of a dedicated Medicare Benefits Schedule item number for ACP remains unresolved in the literature (Pereira-Salgado and Watts 2017) and there was no clear consensus on this issue from our participants. Although the Medicare fees for existing health assessments do not have a strong evidence base (Scott and Connelly 2011), our participants found that health assessments provide time and space to subtly prompt patients to think about ACP while discussing other preventive care issues. Currently, Medicare-funded health assessments are only available to certain segments of the general population, such as those aged 45–49 years or ≥75 years. This misses ACP opportunities for those >65 years of age who are already at risk of developing dementia (Australian Institute of Health and Welfare 2018). Expanding the existing criteria to include periodic health assessments every 5–10 years of life is likely to lead to more ACP before a diagnosis of dementia.

At the level of the practitioner, several of our participants self-identified lack of confidence in assessing capacity as a barrier to ACP in dementia. Therefore, efforts at increasing the uptake of ACP in dementia should focus on improving GP competency in assessing decisional capacity rather than simply focusing on communication skills required for ACP. A Dutch intervention study that involved training GPs in ACP through role-playing workshops demonstrated short-term improvements in ACP for people with dementia (Tilburgs et al. 2020), and this may serve as a template for a similar project in Australia.

This exploratory qualitative study reports findings arising from interviews with a small number of GPs. Our findings may not reflect the experience of GPs in all areas of metropolitan Melbourne, or GPs in rural or remote Australia, so care should be taken in generalising these findings to other contexts. Although we set out to understand how ACP works for patients with dementia, some of our findings are not dementia specific. This is likely due to an overlap in the approach to ACP for people with and without dementia.

The RACGP recognises ACP as the embodiment of person-centred health care and promotes its incorporation into routine care (RACGP 2012). As people live longer, we can expect ACP to become an increasingly important aspect of healthy aging. Culturally diverse public health campaigns may prime more patients to engage in conversations about future care. At the practitioner level, education for capacity assessments may improve GP confidence in dementia ACP. Future studies may investigate ACP for GPs based in regional, rural and remote areas, explore the insights of practice nurses and investigate the patient and carer perspectives of minority populations.

The data supporting this study will be shared on reasonable request to the corresponding author.

The authors declare no conflicts of interest.

This research project was supported by the Royal Australian College of General Practitioners with funding from the Australian Government under the Australian General Practice Training Program.