Prognostication As an Interactionally Delicate Matter: A Conversation Analytic Study of Hospice Multidisciplinary Team Meetings

ABSTRACT

Prognostication has been found to be a delicate matter in interactions between palliative care professionals and patients. Studies have investigated how these discussions are managed and how speakers orient to their delicate nature. However, the degree to which prognostication is a delicate matter in discussions between palliative care professionals themselves has yet to be investigated. This study explored how hospice multidisciplinary team (MDT) members oriented to the delicacy of prognostication during their meetings. Video-recordings of 24 hospice MDT meetings were transcribed and analyzed using Conversation Analysis. In-depth analysis of the interactions showed how prognostic discussions were oriented to as delicate. This was displayed through markers such as pauses and self-repair organization including cutting off words and restarts, and through accounts accompanying the prognosis. In this way, it was seen that prognostication was not necessarily straightforward. This was further evidenced when prognostic requests were problematic to respond to. It is noteworthy that prognostic discussions are delicate during hospice MDT meetings. Potential reasons may reach further than the taboo of death and lie within prognostic uncertainty and accountability. Research is warranted to explore what causes this delicacy and whether specific support is needed for hospice staff.

Introduction

Death and dying are often treated as cultural taboos: difficult or delicate to discuss (Holt, 1993; Holt, 1993). As a result it has been recommended that conversations and stories about everyday death and dying become more common within society (Sallnow et al., 2022). However, this delicacy has also been observed in palliative care settings, where professionals are used to, and skilled in, working with terminally ill patients who are at or approaching the end-of-life. Several studies have shown how clinicians employ different communicative strategies that orient to the delicate matter of these discussions (Ekberg et al., 2021; Parry et al., 2014; Pino & Parry, 2019; Pino et al., 2016).

Despite their taboo nature, topics are not necessarily pre-labeled as delicate in interactions. It is speakers themselves that constate something as delicate through the specific interactive practices used in introducing, pursuing, and closing the business at hand in the interaction (Yu & Wu, 2015). In this way, delicacy is something that is locally oriented to and managed by participants (Silverman, 1997). Therefore, there is a need for closely examining interactions to uncover when and how the speakers treat certain topics, actions etc. as delicate.

Different specific strategies, such as avoidance and depersonalization, have been identified to navigate delicacy (Weijts et al., 1993). This has been labeled as “expressive caution” (Silverman, 1997). Expressive caution can mark delicate objects (e.g., words and phrases) through delay, various speech perturbations, and elaborations and story-prefaces to mark and manage these delicate items (Silverman, 1997). Delay can, for example, be a pause in the interaction occurring before a delicate term (Silverman & Peräkylä, 1990) or through a hesitation marker such as uhm (Lerner, 2013). Pauses or silences are often seen as indicators of interactional trouble due to turns normally being produced without gaps and overlaps in order to minimize silence between turns (Sacks et al., 1974). Another way of displaying orientation toward delicacy has been found during invitations through indirect formulations (Traverso et al., 2018). Certain expressions are known for indicating interactional trouble such as well-prefaced responses (Lerner & Kitzinger, 2019). Laughter can also be used to display delicate orientations to potentially embarrassing or troublesome issues (Beach & Prickett, 2017). Lerner (2013) provides a brief overview of other strategies such as explicitly attending to the delicacy of an expression, whispering, or using a softer voice, and euphemistic formulations. Self-repair, when the speaker deals with some kind interactional trouble in their own talk, can also indicate delicacy in interaction, where conversational repair generally orients to potential issues with speaking, hearing, and understanding talk (Schegloff et al., 1977).

When considering interactional delicacy, it can also be relevant to look at accounts. Providing an account is the practice of explaining the reasons for performing some or other action. In interaction, situations are often encountered in which participants are doing accountable actions in need of explanation, clarification, or justification (Nielsen, 2009). However, it is important to consider when and thereby what people account for, and what that conveys about the interaction. By looking at when a speaker provides an account, it can potentially be identified when something is not entirely straightforward – when something yields to be handled in a way that requires explanation, clarification, or justification.

Prognostication is central to clinical decision-making (Gill, 2012; Hui, 2015). Within palliative care, estimating patients’ length of survival can guide clinicians to use relevant care pathways, and patients might be granted access to certain benefits based on this information (Chu et al., 2019). Although prognostication can involve more than life or death, predictions of life expectancy are what most clinicians consider when they hear the word prognosis (Glare & Sinclair, 2008). In this way, death and prognosis often occur together within palliative care settings. When considering talk about prognosis and death, the evidence mostly focuses on how professionals talk with patients or their relatives (Parry et al., 2014; Pino & Parry, 2019; Wu & Zhang, 2024). Thus, there is a paucity of evidence on how palliative care professionals themselves discuss patients’ prognoses (Bruun et al., 2022), for example in hospice team meetings, with only a few studies having shed light on this area (Bruun, 2023; Bruun et al., 2024). Hospice team interactions may differ from those between professionals and patients and their relatives. The latter are often characterized or looked at as cases of “bad news” delivery (see Maynard, 2003).

The hospice multidisciplinary team (MDT) is essential for providing holistic palliative care (Vissers et al., 2013), and teams should ideally meet weekly to review patients’ care plans (Payne et al., 2022). Recommendations also state that the MDT should be consulted to determine the prognoses of palliative care patients (Chu et al., 2020; Maltoni et al., 2005; National Institute for Health and Care Excellence, 2015). Hospice staff members are therefore used to dealing with and discussing death and dying as this is an essential part of working in a hospice. What remains unexplored is whether there is an orientation to prognostication being a delicate matter during hospice MDT meetings.

Study aim

The study aim was to explore how hospice staff members displayed an orientation to prognostication as being a delicate interactional matter during MDT meetings.

Materials and methods

The study aim was addressed by collecting and analyzing video-recordings of hospice MDT meetings using Conversation Analysis (CA). CA is used to systematically analyze social interaction through close investigation of how participants produce turns at talk (Stivers & Sidnell, 2012). The analyses describe the interactional structure in terms of how practices, actions, and activities are organized by and between speakers. CA methodological tools include recordings of naturally occurring interactions and detailed transcriptions of these data.

A study protocol was developed, and the final version was registered with the Open Science Framework (OSF) on June 4 2021 (https://osf.io/bdf3t). The study was part of a wider project exploring prognostic decision-making of imminently dying patients within specialist palliative care MDTs (Bruun, 2023).

Study setting and participants

Data were collected from a UK hospice providing services for patients with advanced life-limiting diseases and consisting of an inpatient unit, day care and outpatient facilities. The inpatient unit comprised two wards with 15 beds each. A 1-hour MDT meeting was held once a week for each ward and for the hospice outpatients. The MDT comprised a variety of hospice staff (e.g., doctors, nurses, physiotherapists,’ and social workers). The purpose of the MDT meetings was for staff members to discuss and plan the care of patients. Patients themselves did not attend these meetings. All staff who attended meetings during the data collection period, and were willing to provide informed consent, were eligible for the study.

Ethical considerations

The study received a favorable opinion by the London – Camden & Kings Cross Research Ethics Committee (REC) (IRAS Project ID: 276367; REC Reference Number: 20/LO/1168) on December 4 2020. Patients’ consent was not obtained for this study, and for this reason, support from the Confidentiality Advisory Group (CAG) was obtained (CAG reference: 20/CAG/0141) on April 6 2021.

As the meetings involved many varying staff members and staff joining the meetings at different times, it was not possible to inform all attendees about the study in one information meeting, or to inform and obtain consent just before staff entered the meeting room as initially planned. Therefore, meeting attendees were asked to read study information sheets and sign a consent form either before or after attending the meeting.

Consent was provided by 65 attendees. If a staff member did not consent to participate in the study, the meeting was still recorded but their data were not used for analysis. This is the same process described in similar studies recording MDT meetings (Nic a Bháird, 2015; Raine et al., 2014). All study procedures were approved by the REC, including external review of the study protocol from a senior researcher with experience in conducting CA research and collecting data in palliative care settings.

Data collection

Data were collected from May to December 2021. During this period, only one hospice ward was operating due to COVID-19 restrictions. Two cameras and an audio-recorder were used for data collection. The researcher was present during meetings as an observer.

Video-recordings of 24 MDT meetings were collected, yielding approximately 24 hours of data. Each meeting involved 10–15 attendees.

Data management and analysis

Recordings were audibly masked by removing all participant identifying information (i.e., patient and staff members’ names and locations) before being securely stored. Sequences of talk involving prognostication were identified by one researcher and transcribed following standard CA conventions (see Supplementary File 1) (Hepburn & Bolden, 2012; Jefferson, 2004).

CLAN software version April 28 2021 or above was used to support the transcription process. Single-case analyses (Pomerantz & Fehr, 1997) were conducted to create a collection of cases (i.e., collection analysis). Collection analyses systematically explore patterns of an interactional phenomenon (Hoey & Kendrick, 2017). Data and analyses were discussed in CA data sessions; a common practice within CA (Stevanovic & Weiste, 2017).

Results

Different features of hospice staff members’ talk displayed an orientation to prognostication as being a delicate matter. The data revealed that certain markers were present, which included pauses and self-repair organization such as cutting off words and restarts. Prognostic utterances also occurred with statement-assessment sequences accounting for the prognosis. These all indicated that prognostication was not a straightforward action.

Each of these markers are described in the sections below. It should be noted that the features presented above overlap a great deal. In this way, there will be delicacy markers, accounts, and displays of non-straightforwardness in each of the analyses. Each excerpt will focus on one feature in order to highlight them individually. A speaker acronym key is provided in Supplementary File 2.

Delicacy markers

Delicacy markers included pauses and self-repair organization such as cutting off words and restarts when hospice staff provided prognostic utterances.

In Excerpt 1, staff members are discussing a patient whose mental capacity they have queried. This has led the team to plan for the patient to undergo a capacity assessment to clarify whether Deprivation of Liberty Safeguards (DoLS) involvement is necessary. The social worker has confirmed that they will conduct a capacity assessment after the MDT meeting.

Excerpt 1. She’s a dying woman (2021.08.12).

Just before the excerpt, a nurse (NUR) is presenting what she has experienced when caring for the patient over the last couple of days.

In the opening lines (01–06), the nurse is describing the patient’s abnormal behavior involving confusion, asking for alcohol, and swearing. In line 13, the doctor (DR2) initiates an utterance with a cut-off phrase I me-. The doctor then revises this phrase with a prolonged my: and a short pause. The prolongation of my: may be a way for the doctor to hold the interactional floor by signaling that the speaking turn has not yet been completed. She then produces my impression which would be expected to project a verb such as is. However, she abandons this turn progression and provides three cut-off Is followed by the verb feel. Here the doctor is presenting a feeling about the DoLS, but this project ends early with another incomplete sentence; a dols might. After another pause, she revises her utterance and says that she thinks the patient is dying. This time the doctor is using the verb think which implies an opinion or thought. In line 15, another pause occurs, and then she states that the patient is dying imminently. The word dying is said with a creaky voice and the rest of the utterance is produced with low speech volume. The utterance ends with an and, which, as a conjunction, is highly projective. The doctor continues in line 20, where she, after a restart, provides a sequence comprising several statements and assessments of the patient (i.e., statement-assessment sequence) that accounts for her prognostic utterance in line 16.

In this excerpt, the doctor ended with a rather strong prognostic utterance in line 16, however before arriving at this utterance, there had been several markers of interactional trouble. There were multiple occurrences of self-repair such as cutting off words, restarting utterances and several pauses. There were also other indicators that oriented to interactional trouble such as the shifting between formulations that related to the doctor’s assessment of the patient (my impression, I feel, and I think) and low speech volume. Lastly, the doctor ended with a statement-assessment sequence accounting for the prognosis. In this way, there were several markers or indicators of interactional delicacy when providing the patient’s prognosis.

Accounts

The data also revealed that prognoses were formulated with additional accounting information. This was accomplished through statements and assessments of the patient’s state. This led to the introduction of the statement-assessment sequence. The statement-assessment sequence comprised statements about and/or assessments of the patient where several observations, symptoms, assessments and sometimes interventions were listed, collating information about the patient’s state. These sequences could function as accounts when occurring either before or after a prognostic utterance.

In Excerpt 2, the nurse (NUR) begins her case presentation of the next patient for discussion at the meeting.

Excerpt 2. She’s really taken a dip (2021.17.11).

The nurse begins the presentation by mentioning the patient’s name, age, length of admission, and the patient’s What Matters to Me (i.e., a meeting template item, used to capture what is important to the patient on the day of the meeting). In line 09, she produces a prognostic utterance, where she states that the record of the patient’s Phase of Illness (i.e., a tool used in advanced illness to describe distinct stages of an individual’s illness according to their care needs (Mather et al., 2018)) is not “deteriorating,” where she asks the team whether they should say “dying” and thereby change the phase. The doctor (DR1) confirms and agrees with the nurse through nodding and her yeah in lines 11–12. The nurse then mentions that the patient’s decline started the evening before, and that her condition has got worse overnight. Here, the nurse initiates a statement-assessment sequence, where she shares her observations and assessments of the patient. This sequence seems to account for why they should change the Phase of Illness. She continues by mentioning that they have started the patient on a syringe driver (line 19). This leads to a second upgraded prognostic utterance, where she says that she thinks the patient has a small number (smatter) of days left to live. Her prognosis receives confirmation from the doctor in lines 24 and 26. The nurse then provides another short statement-assessment sequence where she says that the patient is barely talking and cannot hear the staff, which implies that the patient has indeed deteriorated. This seems to provide evidence of the patient’s poor state and thereby account for the aforementioned prognosis. Lastly, this leads to a proposal of changing the record of the patient’s Phase of Illness (dying) and Karnofsky Performance Status (KPS) (Schag et al., 1984) score (twenty) in line 30, which the doctor confirms. After the excerpt, the nurse continues with the patient presentation and lists the patient’s diagnosis and main issues according to the MDT meeting template.

In this excerpt, there were several occasions where the nurse accounted for her prognostic utterance. The nurse provided a statement-assessment sequence after having proposed that they should change the record of the patient’s Phase of Illness. After providing her second prognostic utterance, the nurse again accounted for the prognosis through another statement-assessment sequence. This displayed an orientation to and treatment of prognosis as something that needed to be explained or justified in (clinical) evidence. Here, the evidence was the nurse’s observations and assessments. This became even more clear, when the nurse simply could have continued with the update of the Phase of Illness and KPS score (as seen at the end of the excerpt) after the doctor’s confirmation. Instead, she accounted for the change and her proposed prognosis. Returning to the update of the record after having provided several justifications for the change displayed that the nurse treated this as something needing to be accounted for.

This section showed how participants displayed an orientation to prognosis as something needing to be based on clinical evidence. It was shown how evidence such as observations and assessments were used to justify and therefore account for the provided prognosis. The evidence was provided through the statement-assessment sequence with multiple utterances describing the patient’s state.

Prognostic request as non-straightforward

So far it has been shown how certain markers and accounts display an orientation to prognostication as being a delicate action. As noted earlier, this implies that prognostication is a non-straightforward practice. This notion of non-straightforwardness will be further explored in the analysis of Excerpt 3. This excerpt will show that a request for prognostic information is not straightforward for a doctor to respond to where the type-conforming response (Raymond, 2003) to the prognostic request is delayed and several delicacy markers and accounts occurred.

Excerpt 3. How long do we think the decline’s gonna be (2021.15.12).

In the excerpt below, a patient’s daughter has expressed concerns about discharging her mother from the hospice. This has led the MDT to discuss the patient’s current state, where they seem to agree that there has been a decline.

In the first lines (01–03), the ward manager (WAR) provides an explicit request for the patient’s prognosis through the question; how long do we think the decline’s gonna be. One of the doctors (DR1) then produces an utterance in lines 04–06. This utterance begins with two cut-off words followed by a pause and a prolonged u:hm. After another pause, she makes a click sound and produces another cut-off word. She then states, after several cut-off words, that the patient’s liver is full of mets (metastases). However, before completing this utterance, several pauses and more cut-off words occur. With this statement of the patient’s liver, it becomes clear, that the doctor is initiating a statement-assessment sequence of the patient’s current state. The doctor then continues by referring to something about the patient in line 08, but the utterance appears incomplete with the cut-off words in line 10. This project is abandoned, and instead the doctor states that the patient is in bed and continues by explaining that the patient only gets up for a very limited time during the day. Other team members agree with the doctor’s statement. Through these statements, it also becomes more apparent that the ward manager’s question is not straightforward for the doctor to answer. The type-conforming response (Raymond, 2003) to her question would be some kind of timeframe relating to how long the patient’s decline is going to be. Instead, the doctor provides a non-type-conforming response with an assessment or statement of the patient being close to bedbound.

The doctor’s statement leads to another doctor (DR2) mentioning that the patient is poorly (line 26), which receives affiliative responses in the following lines. The first doctor (DR1) then initiates a new utterance that begins with I and a cut-off word. She then says the words although, I don’t and I imagine, where the doctor restarts the utterance several times. The utterance ends with the doctor’s prognostic utterance in line 31, where she says they are looking at a prognosis of short weeks. In this way, the answer to the ward manager’s question in lines 01–03 is provided several turns later, not until line 31. The other doctor (DR2) agrees with this prognosis in line 32. Then the first doctor (DR1) provides an account for this prognosis by saying that it is based on the disease trajectory. After this excerpt, the doctor further accounts for the prognosis by explaining some of the patient’s blood results.

This excerpt showed how several markers of interactional trouble or delicacy occurred. There were multiple (long) pauses, cut-offs, and restarts indicating a lack of straightforwardness in the interaction. The non-straightforward aspect was also seen through the use of accounts. The actual type-conforming answer to the prognostic request occurred several turns later in the interaction, after the doctor had produced a statement of the patient being close to bedbound. After responding to the request, further accounting from the doctor occurred as well.

Discussion

This study showed how hospice MDT members oriented to prognostication as a delicate interactional matter. Prognostication appeared to be a non-straightforward action, which was displayed through markers indicating interactional trouble such as pauses, word cut-offs, restarts, and rather significant accounts when staff members provided prognoses.

These findings are in line with other conversation analytic studies showing that discussing death is a delicate matter (Ekberg et al., 2021; Holt, 1993; Parry et al., 2014; Pino & Parry, 2019; Pino et al., 2016). As noted previously, the hospice MDT meeting is a different interactional situation compared to those in which professionals are talking about death with patients and/or their relatives and from everyday conversations. In the MDT meeting, participants do not have to deal with the emotional aspect of communicating bad news to patients or their next of kin. Instead, the hospice MDT meeting is an institutional meeting between professionals designed to discuss their organizational tasks and ”clients” that often involves talk about prognoses and death. Death and dying is a common topic for these professionals since the hospice is caring for terminally ill patients who are close to death. It is, or becomes, everyday routine for staff to deal with death and dying. For this reason, it is noteworthy that there is still a display of delicacy present when discussing prognosis.

Discussions around patients dying and their prognoses have been found to be emotionally difficult or hard for staff (Mack & Smith, 2012). Despite hospice staff being used to dealing with death and dying, the emotional labor and impact of a patient’s poor prognosis on the professionals themselves might explain why there is still a delicacy around these discussions. It might be less emotional to discuss death and prognosis with the MDT, but nonetheless hospice work remains emotionally challenging for professionals by having to care for the dying, supporting patients’ relatives (see Funk et al., 2018), and also dealing with their own emotions. It has been observed how palliative care MDT meetings serve as a place for staff where they can express their emotional responses to patient cases (Borgstrom et al., 2021). Further research into the emotional burden of prognostication during MDT meetings is needed.

There is an additional layer to consider when dealing with prognostication: providing an accurate prognosis. Instead of the delicacy orienting toward death per se, it might also orient toward uncertainty and difficulty with attempting to predict length of survival, particularly in a group environment. Evidence suggests that the hospice MDT uses rather unspecific references to time when providing prognoses (Bruun et al., 2024). When providing specific prognoses, professionals can be held accountable for their predictions. Prognostic uncertainty has also been listed as a reason for clinicians avoiding prognostic discussions with patients (Travers & Taylor, 2016).

Other studies have shown how clinicians navigate prognostic uncertainty in conversations with patients and their next of kin (Anderson et al., 2020). However, clinicians might be held accountable in a different way compared to when providing prognoses to patients and their relatives. During the MDT meeting, and in the hospice as a workplace, there is a professional reputation to maintain and there is a risk of jeopardizing that by providing an inaccurate prognosis. In this way, the one providing the prognosis could be held (negatively) accountable by their colleagues. Here, there could be substantial face-work for the professionals to maintain, relating to an individual’s image of self in terms of approved social attributes from others (Goffman, 1955). By providing a wrong or inaccurate prognosis, a staff member’s self-image can be challenged if colleagues hold the person negatively accountable for it. However, one study showed that doctors tend not to hold colleagues accountable for prognostic errors (Christakis & Iwashyna, 1998). The analyses presented in this paper did not deal with professional reputation or facework. As a result, the role uncertainty and accountability play in these interactions would benefit from further exploration.

As recommendations state that the hospice MDT should be consulted when providing patient prognoses (Chu et al., 2020; Maltoni et al., 2005; National Institute for Health and Care Excellence, 2015), future research should further explore the sensitivity of prognostication within the hospice MDT and its potential impact on aspects such as meeting length, prognostic accuracy, and patient care. Studies into the perspectives of staff on prognostication are warranted to gain a deeper understanding of the potential facilitators and barriers as to why. Such research could also inform the understanding of whether support is needed for the MDT regarding prognostication.

The study findings add to the sparse evidence on how hospice MDTs carry out prognostic communication in clinical practice. A robust evidence base is needed to effectively improve prognostic communication as it is necessary to base any intervention or recommendation on how MDTs currently communicate. Evaluating and reflecting on current practice would be the first step to potentially change practice and identify good ways of doing MDT prognostication.

Study strengths and limitations

Through detailed analyses of the moment-by-moment interaction as it naturally occurred during hospice MDT meetings, the analyses shed light on the delicacy of prognostication. The use of video-recordings ensured that findings were based on real interactions from clinical practice. The dataset comprised 24 meeting recordings, which was enough to identify different practices in the interaction. Data and preliminary analyses were discussed in data sessions ensuring transcription accuracy and validation of findings.

All hospice staff members and visitors had to wear face coverings when entering the hospice due to COVID-19 restrictions. This meant that in cases where it was not entirely sure who was speaking, participants’ mouth movements could not be used for clarification. This meant that transcription often relied on recognizing participants’ voices and making (reasonable) assumptions about who was speaking. Poor sound quality due to background noise also sometimes made it difficult to hear and therefore transcribe what participants said. Despite having three devices capturing sound from different positions, it was sometimes not possible to hear and determine what the speaker said. This explains the (frequent) use of brackets in transcriptions that propose a possible interpretation.

It could be perceived as a limitation that the researcher was present during the meetings. Because of this, it was not possible to completely eliminate researcher influence on the interactions. However, it has been argued that “researcher-participants do not (necessarily) challenge the local ‘naturalness’ of the data” (Hofstetter, 2021, p. 2), and that researcher participation can be useful for fieldwork providing evidence for the researcher’s unique adequacy and for gaining access to the activity (Hofstetter, 2021). Visitors and/or observers commonly participated in these meetings, and thus presence of visitors was not regarded as being unusual. The role of the observer’s paradox (see Labov, 1972) should also be considered and whether staff altered their language or avoided expressing certain (potentially sensitive) matters that they did not want captured in the recordings.

Conclusion

Prognostication was seen as interactionally delicate during hospice MDT meetings. This was displayed through delicacy markers including pauses and self-repair organization such as cutting off words and restarts. Prognoses were accounted for with statements and assessments of the patient’s state where several observations and assessments were listed. This showed how prognostication was not straightforward in the interaction, which was further evidenced when responses to prognostic requests were delayed. The interactional delicacy may be grounded in prognostic uncertainty and potential accountability. Future research is warranted to explore what causes this delicacy and whether further support is needed for hospice staff.

Acknowledgements

The authors would like to thank the collaborating hospice and all the staff members who took part in the study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/10410236.2024.2380959

Data availability statement

Due to the sensitive nature of the data, recordings and transcripts cannot be shared.

Additional information

Funding

This work was supported by Marie Curie UK under the Chair’s Grant MCCC-FCH-18-U.