Initiating decision-making conversations in palliative care: an ethnographic discourse analysis

The methodology adopted to answer the research question consists of a combination of discourse analysis with ethnographic observations. A social psychological approach to discourse analysis was selected in order to analyze how decision-making processes are constructed through discourse as well as the social consequences of these processes [ 23, 24]. For the purpose of this research study, “discourse” is defined as all forms of spoken and non-spoken interactions that construct social reality [ 23], the social reality in this case being decisions about palliative care options. Ethnographic methods were also borrowed from organizational ethnography, which is a qualitative methodology involving long-term participant observation in an organization in order to understand its culture [ 25]. During field observations which spanned a year, special attention was paid to the ways in which the context and organization of care influenced the discourse constructing decision-making processes. The methodology aimed to capture the discourse that initiated decision-making conversations during clinical encounters and to examine how this discourse related to the organization of care. This methodological combination represents the best design for obtaining a longitudinal perspective on naturally-occurring discourse during sensitive palliative care interactions as well as for discerning how the organization of care relates to these discourses, taking factors such as long-term doctor-patient relationships into account.

The palliative care providers participating in this study were family physicians and pivot nurses. They were mobile throughout a community acute-care hospital in Montreal, Canada, and also followed patients at the outpatient clinic of the Family Medicine Unit and at home in partnership with a local community health center. The majority of referrals came from the team of oncologists, especially when prognoses worsened and intractable symptoms appeared. However, this study includes referrals from emergency medicine, intensive care, as well as from other family physicians who practiced in the community and did not provide palliative care. The palliative care service also received many referrals from geriatricians, but these patients had cognitive problems that prevented their participation in decision-making and thus precluded their participation in the study. Referrals also occurred whenever patients and families expressed the desire to end active treatments. In 2009, 192 patients being cared for by the team died at the hospital, while 53 patients died at home and 30 patients died after being transferred to a nearby palliative care unit. Patients were generally followed by the same family physician as outpatients or while receiving care at home, unless they experienced an acute episode that needed stabilization as an inpatient, which would be handled by one of the four family physicians on call at the hospital that week. After such an episode, and depending on their condition, patients would either be asked to come back to the outpatient clinic for follow-ups or would receive home visits. The pivot nurses oversaw the continuum of care. If patients and their families decided that death should occur at home, this was arranged with the support of the local community health centers whose nurses performed daily visits and communicated with the palliative care team. If returning home was no longer possible because of medical or social circumstances and unless death was imminent, patients were generally redirected to other services, which included long-term care, or inpatient palliative care units with dedicated beds for when patients’ life expectancy was predictable and limited. There were, however, waiting lists for such services, and it was common for patients to worsen and to die while still under the care of the team in the acute care facility. According to internal statistics, the team estimated that, every year, an average of 40 patients died under their care in the hospital while waiting for a place in a dedicated palliative care unit.

Data were generated over one year of participant observation in the palliative care team (April 2010-2011), where the field researcher (EB) observed interactions from Monday to Friday between 9 a.m. and 5 p.m. at a minimum. Consultations between family physicians and patients recruited for the study were observed across all care settings: the hospital, the outpatient clinic, and at home. Consultations were audio-recorded with the patients’ permission. The field researcher kept detailed field notes of all observations, including consultations, team meetings, informal conversations before and after consultations, and so forth. She was present at all weekly interdisciplinary meetings of the palliative care team and had a desk in the office of the head of the unit, next to the pivot nurses’ office. In addition, she participated in daily rounds with residents when recruited patients were hospitalized and attended the residents’ lectures. A field journal was kept throughout the fieldwork to facilitate reflection during both data gathering and data analysis.

Physicians were responsible for recruiting patients and obtaining permission for the researcher to attend the consultation where official consent forms were signed. Patients were included if they were cognitively able to participate in decisions (according to their treating physicians) and spoke English or French. Recruitment occurred through a purposeful sampling procedure [ 26] by which we sought to recruit new patients under the care of all the different health care providers. The field researcher followed 10 participants at a time in order to reduce scheduling conflicts between observations of inpatient and at-home consultations, which could occur simultaneously when performed by different care providers. She updated the team regarding the need to recruit new patients at the interdisciplinary meetings. A total of 30 patients verbally agreed to participate in the study. However, seven of these patients either deteriorated too rapidly to sign the consent form, or were discharged before signing the consent form and were not readmitted during the duration of the study. Five additional patients were not retained for analysis because there were insufficient data. A total of 18 patients were followed longitudinally throughout the course of their care, and six family physicians as well as two pivot nurses participated in the study (see Table  1 for patient characteristics, origin of referral, and number of consultations observed). Many different types of decisions were observed, including symptom and medication management (pain, anxiety, depression, shortness of breath, nausea, constipation), place of care, radiotherapy, chemotherapy, surgery, antibiotics, do not resuscitate orders, etc.

The field notes contained detailed ethnographic data and allowed us to develop a thorough understanding of the context of decision-making processes and of the organizational culture of the palliative care team. The audio-recorded data were transcribed and decision-making conversations were extracted from the discourse (Please refer to the Appendix for the Transcription Key). Decision-making conversations were defined broadly to include all interactions about given topics regarding therapeutic action such as initiating, modifying or stopping a medication, procedures or therapy. We then proceeded to identify the discursive practices initiating all the decision-making conversations identified, searching for differences and consistencies as well as for their function and consequences based on the linguistic evidence at hand, such as the interlocutor’s reaction [ 23, 24]. Close attention was paid to the types of discursive practices that initiated decision-making conversations, namely the recurring types of questions or arguments that were used to introduce new decisions, and to how these discursive practices related to the organization of care and to contextual features such as the type of decision being introduced, the length of follow-up, and so forth. Data coding involved several iterations as well as the use of HyperRESEARCH 3 qualitative research software. Data were analyzed in their original language; French excerpts were translated into English by the first author and verified by a professional translator. The results show the discursive practices that were found systematically and repeatedly in the data.

The timeliness of referrals to the palliative care service and the origin of these referrals – i.e. whether requested by the patient or by another health care provider – both affected the discursive practices initiating decision-making about palliative care options. The timeliness and origin of referrals had an impact: (1) on whether patients were still conscious and able to participate in decision-making and could thus be recruited into the study; (2) on the decisions that remained to be made; and (3) on patients’ awareness of their terminal status. It was clear that participating family physicians promoted the importance of introducing decisions about palliative care options early on, before patients could no longer participate in such decisions. This was reflected not only in the way they consistently introduced decision-making conversations early on, but also in the way they taught residents about the need to discuss future options of care and the consequences of failing to do so. For example, during a teaching round observed by the field researcher, a physician (MD 2) gave the example of a surgical intervention that was being done to reverse renal failure in an unconscious, hospitalized patient. The attending physician explained to the residents that such an intervention, while preventing imminent death, could also potentially result in further painful complications. Emphasis was put on the fact that had appropriate discussions occurred earlier, the patient would have had the opportunity to refuse this type of intervention. When patients were referred to the palliative care service at a very late stage in the progression of their illness, their ability to participate in such decisions was reduced, and the decisions that remained to be made mainly revolved around terminal symptom management.

In instances where patients requested a follow-up for symptom control or refused active treatments themselves, physicians did not have to justify their presence and decision-making conversations could proceed in a straightforward manner, because patients discussed their death directly. For example, in the context of a consultation in intensive care, a physician said: “I’m here to see you because I’ve been told that you wish to stop receiving your transfusions. What makes you say that you’ve had enough?” The patient replied: “It’s not improving. I receive the transfusions and two days later I am back here. I’m tired just walking from the bed to the armchair. I’m tired talking and lying down in my bed. It’s not going better”. Requesting that active treatments be stopped suggests an acknowledgement of terminal status, which the physician then verified by asking: “And the oncologists have told you that there isn’t any miracle treatment for you?” To which the patient replied: “No.” (Field notes, intensive care consultation, MD 3, Patient 3).

Decisions regarding symptom management were ubiquitous in the data. Decisions about interventions to alleviate symptoms were directly influenced by patients’ feedback in an attempt to maximize comfort. Symptom control concerned patients’ primary complaints, such as pain, nausea, constipation, shortness of breath, and other types of discomfort. These decisions were either introduced by physicians through general opening questions, or by patients who presented complaints. There was a formulaic quality to the initiation of these decision-making conversations, insofar as they followed a traditional history-taking model. Physicians first asked general open-ended questions, and then followed-up with more specific questions. This was the case in Excerpt 1, where the physician asked a question about the breathing difficulties in general and then specifically about any secretions. The discursive practices of both patients and family physicians constructed decisions about symptom management as needing immediate attention. As such, addressing these decisions did not require justifications, and inquiring about any discomfort represented a common and expected role for family physicians.

For family physicians, engaging in decision-making about symptom management also involved ascertaining whether or not certain complaints from patients indeed needed to be addressed. More specifically, in response to patients who brought up different symptoms requiring decision-making, family physicians would first make sure that patients actually wanted the problem fixed, especially if the symptoms in question were not corroborated with non-verbal displays of discomfort (e.g. frowning and restlessness). In Excerpt 2, the physician verified the relative significance of the symptom in relation to the patient’s level of comfort by asking about sleep and whether the stinging pain was bothersome.

Many palliative care decisions, by default, address the timing and cause of death, making them particularly delicate and emotional. During the consultations observed, there were many decisions about therapeutic options that involved comparing outcomes and thus confronting the future. Once the health of patients declined, conversations about the future would invariably underline their limited life expectancy and entail discussing their death. The daily work of the palliative care team thus involved both symptom management and raising decisions such as the preferred setting for end-of-life care.