1.1 CA on narratives in health care interactions

CA is a method to describe patterns of naturally occurring interaction, which can be used to unpick patient-professional communication. Within conversations, speakers are taking turns at talk. They are entitled to a single-turn constructional unit (an utterance) and when that unit reaches a recognisable point of completion the speech initiative may alter between participants.²² Narratives and stories often stretch over longer sequences, consisting of multiple-turn constructional units, which requires specific and subtle negotiations of speech initiative that may differ depending on whether the narrative is initiated by the speaker or solicited by the recipient.²² CA has been applied to study activities in health care communication directly or indirectly related to the production of patients' narratives.^23-26 Findings relevant for conducting PCC are, for example, how access to the first- and third-turn slots authorizes HCPs to introduce topics and be in charge of the agenda,^{27, 28} and how patients nevertheless can find strategies to share their perspectives by going beyond the topical agenda.²⁴ Patients' narratives are also shaped by the way HCPs pose questions: experience-oriented, affective or general-inquiry questions generally support narrative production,^{23, 24, 29} as does experience-oriented formulations²⁴ and tentative speech.²³ Drew et al.²⁶ and Kettunen et al.²³ have also described the importance of leaving conversational space open in order for patients to contribute with their point of view, something which may be undermined if the conversation is too heavily standardized.²⁶

However, with the exception of Drew et al.,²⁶ none of these studies is conducted within the inferential framework of PCC, which constitutes a particular kind of institutional encounter by placing the patients' narrated experiences of their condition as the primary basis of care and treatment planning.⁸ Our starting point of this paper is that the normative appeal to practise PCC serves as a particular context for framing the interaction between HCPs and patients, and that narrative elicitation, especially when serving as a bannister for the realisation of the ethics of PCC in a microperspective, warrants further scrutiny and exploration. Our aim is to explore what activities are performed and what challenges participants face in the interactive process of narrative elicitation, carried through in patient-professional communication as part of a remote PCC intervention.

2.1 The intervention study

The present study is part of a larger randomized controlled trial (RCT) evaluating the effects of remote PCC as additional support, complementary to usual care, for patients on sick leave due to common mental disorders (CMDs).³⁰ The RCT took place in an urban region in Sweden, and patients were recruited from nine primary health care centres. The intervention was administered remotely, through access to an interactive digital platform and through phone support with HCPs trained in PCC. Although the content of the interactive platform was adapted to suit the conditions of patients included in the study (mild to moderate CMD), the primary aim of the intervention was to create a partnership with the patient, based on their narratives, to support them in achieving what was identified as their health-related goals. According to the evidence-based PCC framework^{8, 13} used in the study, this is partly achieved through the narrative approach of the PCC phone conversations. The HCPs who worked with the intervention had varying degrees of experience in the theory and practice of PCC, and took part in regular meetings with experienced clinicians and researchers for continuous reflection and training in the ethical principles and practice of PCC. Part of this training consisted of reviewing some of each other's conversations and giving each other feedback on conducting PCC over the phone. None of the HCPs had any prior experience of working specifically with CMDs but participated in a half-day lecture led by experts in CMDs. As the intervention was performed in addition to usual care, the patients were managed in parallel at their primary health care centre according to guideline-directed care. When patients were allocated to the intervention, a phone conversation with the HCPs was scheduled to occur within a week or two. Unless the patient declined, these phone conversations were recorded, and these audio recordings are the material used for the present study.

2.2 Data collection

Data for the present study were collected between the start of the RCT in February 2018 and November 2018. During this period, three HCPs (two registered nurses [RNs] and one physiotherapist [PT]) worked with the RCT and a total of 22 initial phone conversations were recorded. Only initial conversations were included because they represented the first occasion for narrative elicitation. One recording was excluded due to sound disturbances. More than half of the remaining audio-recorded conversations were conducted by one of the RNs (called RNa in the excerpts). To minimize the risk of a biased sample dominated by conversations led by the same HCP, we eliminated, at random, some of the recordings conducted by RNa. This resulted in a final sample of eleven audio-recorded conversations, fairly evenly distributed between the three HCP (five recordings by RNa, three by RNb and three by the PT). All three HCPs were women, five of the patients were men and six were women. The length of the conversations varied from 23 to 62 min, and the 11 audio recordings amounted to a total of 462 min. Ethical approval was obtained from the Regional Ethical Review Board in Gothenburg, Sweden (DNr 497-17, T023-18 and T526-18). Participants gave written consent to participation, and oral consent was given before starting recording at the beginning of each new conversation. The transcripts were treated with full confidentiality and names or identifiable characteristics have been altered.

2.3 Analytic procedure

The first author transcribed the conversations with attention to interaction details suitable for the present analysis, for example, emphasis, notations of pauses and micro-pauses and overlapping speech (see the Supporting Information Appendix for transcript legend). The conversations were conducted in Swedish, and so was the transcription and analysis. Thereafter the analysis, as well as the excerpts from the transcripts, were translated into English. All recordings and their respective transcripts were examined by the first, second and last author, using CA.

The analysis was oriented to how narratives were achieved, how they were accomplished in the interaction and managed turn-by-turn.^{22, 31} First, narrative sequences in the conversations were identified and an initial grasp of their content and interactional form was sought. The narrative sequences were then discussed in terms of their location in relation to the overall phases of the conversations. Then, each phase across the 11 conversations was analysed separately to search for interactive patterns in how narratives were introduced, if they were elicited by the HCP or self-initiated, how narrating occurred (or the lack thereof when sought) and what type of activity the narrative produced. Interactional patterns from all 11 conversations were then identified and analysed in terms of differences and similarities. Our analytical focus was thereafter directed to the opening sequences and the main project of the conversations. Excerpts including three different patterns were chosen to scrutinize the preliminary analysis in a data session with interaction scholars from other research fields. The results were then discussed among all coauthors, providing a further examination of the trustworthiness of the findings.

3.1 Opening sequences

Overall, the conversations followed a general structure of an opening phase, followed by the main project of narrative elicitation, a summary phase and a closing phase. The opening phase included greetings, and in half of the conversations, there was a preparatory sequence explaining, for example, the aim of the conversation (its open-endedness) and role expectations (patients choosing topics they find relevant). Excerpt 1 (Table 1) illustrates one such preparatory sequence (see transcript legend in the Supporting Information Appendix).

In Excerpt 1, the HCP explains the intention of the conversation. In Line 7, the HCP begins her utterance with ‘actually’, framing what she is about to say as something that might be contrary to what is expected. She then goes on to explain that this conversation is supposed to be about matters that are meaningful to the patient, who can choose what to talk about. Next, the HCP clarifies how she will participate during the conversation, making it clear that the patient should be ‘running it’ (Line 15).

When a transfer from the opening phase to the main project occurred, this was explicitly signalled by the HCPs. Often, they used a ‘please go ahead’ or ‘where should we start,’ followed by an open-ended question either directed retrospectively at what had happened, or to the patient's present status or circumstances. The HCPs could also suggest several possible topics for the patient to choose.

3.2 The main project: Shaping the conversations in narrative form

Narratives could occur by self-initiation in any of the conversational phases, but their elicitation was at the core of the main project. Below, we will show patterns of how narrative elicitation was carried out in our material and the different kinds of interaction from which narratives emerged. There are patient-driven narrative sequences, question-driven narrative sequences and co-constructed narrative sequences (Table 2). These patterns of interaction should not be understood as describing the entire conversations but only elicitation sequences within the conversations. One conversation could thus include sequences of patient-driven narration, followed by a sequence of co-construction, and so forth.

3.3 Patient-driven narrative sequences

Patient-driven narrative sequences are sequences in the conversations where the patient held the initiative in telling a narrative. The narratives produced are ‘story-like’ in terms of giving insight into a line of inner and outer events. The excerpt below (Table 3) is an example of such a sequence. Here, the HCP hands the floor over to the patient and thereafter holds a nondirective role. While the patient tells her story, the HCP remains in the background, demonstrating her ongoing attention by frequent mhm's, but not interfering with the patient's narration (see transcript legend in the Supporting Information Appendix).

In Excerpt 2 (Table 3), the HCP solicits a narrative by introducing possible topics for the patient as a starting point (Line 3). The patient accepts the invitation to narrate (Line 7), sets the frame for her story by stating that she will ‘start at the beginning’, and hints at the entrance point for her storyline (Line 9). She then tells her story in a chronological fashion until returning to the present time (Line 30), indicating the completion of this particular narrative sequence. Through the narrative of how it all started a couple of months ago, the patient describes how she experienced symptoms (Lines 13, 19, 22) and the events linked to her deteriorating psychological condition (line 15+ in the omitted turns). In doing so she also produces an account that makes her current sick leave understandable, and justifiable. The narrative, which is delivered without any gaps in the interaction flow, comes across as complete—a story acknowledged by the HCP without any requests for clarification or elaboration. Every HCP turn throughout the sequence is used for listener tokens, allowing the patient to tell her story without interrupting. When the patient wraps up, by returning to present time circumstances, the HCP aligns with the previous turn (Line 35), by directing her question towards the present rather than to earlier, uncommented, parts of the sequence.

3.4 Question-driven narrative sequences

In question-driven narrative sequences, the project of narrative elicitation is not carried out smoothly. The participants seem to be struggling with the contributions of the other and they lack any extended turns of talk. The excerpt below (Table 4) illustrates one such sequence in the material. Overall, this was the least frequent pattern of narrative elicitation efforts. Other similar sequences were shorter and typically resolved sooner.

The sequence in Excerpt 3 illustrates the first minutes of the conversation and is only preceded by a short sequence of greetings. Throughout the sequence, interaction is problematic. The professional makes several attempts to shift the speaker initiative from herself to the patient, but these are not successful. The HCP's first question, ‘How are you doing’, is an open, neutral, quotidian question, but there is an element of ambiguity as it can be understood either as a common greeting or, given the health care context, as a request for information. The ambiguity of the previous turn remains as the patient hesitates (indicated in the excerpt by the micropauses) and gives a vague and minimum reply to how he is feeling this very day. The follow up from the HCP, a request for the patient to elaborate on his answer by describing what ‘pretty good’ is like (Line 3), receives a dispreferred response.²⁷ The patient hesitates multiple times, exhales loudly, and struggles to answer the request (Lines 4–11). Already in Line 11, he wraps up his answer by summarizing his status (overall I feel pretty good) followed by a problem formulation (I've been sleeping very poorly so I'm tired), which could be read as an account of his lack of elaboration, or alternatively, an initiative to change the topic. In her following turn, the HCP seeks clarification on the extent of the sleeping problem (you mean last night) and the patient's short, confirmative reply closes the topic. On Line 17, the HCP instead returns to the previous topic of the patient's current status and requests an elaboration by suggesting a comparison of how he is doing now with how he has been doing earlier. The patient again acknowledges the request in the previous turn, but in a minimally informative way. This pattern continues throughout the sequence, giving the overall impression that the patient understands the HCP's requests, and his minimalistic, but accurate, replies can be understood as a compromise not exposing more of himself than necessary. Respecting another person's integrity thus becomes a delicate matter when pursuing to understand more about a person who is restrictive in the information they may want to give as a patient. In sum, the question-driven narrative sequences are characterized by the patient not taking on the role of narrator and repeatedly handing the speech initiative back to the HCP. In contrast to the patient-driven narrative sequences, the question-driven sequences require more coaxing from the HCP. Still, the interaction takes on a more conventional question–response pattern.

3.5 Co-constructed narrative sequences

In co-constructed narrative sequences, the narrative is treated as more of a collaborative project. This pattern will be illustrated by three excerpts from two different conversations. The first two excerpts display the joint elaborations of a narrative displaying gaps and the third excerpt shows how a narrative becomes a means for a joint reflective process.

3.6 Jointly elaborating the content of a narrative

In the excerpt below (Table 5), the HCP elicits a narrative around the circumstances preceding the patient's sick leave. Through her narrative, the patient discloses that this is an emotionally challenging subject.

In Excerpt 4, the HCP elicits a narrative about what happened that caused the patient to take sick leave (Lines 1–3). She starts to describe the situation at her current job, explaining some related circumstances that contributed to her reaching a tipping point (Lines 4–24). Throughout these turns, her frequent use of epistemic markers such as ‘probably’ and ‘I guess’ reveals uncertainty as to what happened.³² In Line 24, she stops her sentence short, as she starts crying. The narrative the patient begins telling in this sequence conveys a vague story of failure, which is charged with emotions, and throughout this sequence, the HCP does not intervene much. The most prominent participation from the HCP is her frequent use of listener tokens (mm-hm), signalling that she is paying attention and that the patient may continue speaking. In Lines 44–45, after an utterance that could be interpreted as a story exit (Line 43), she requests clarification of the conditions of the patient's employment. The patient responds by further explaining the circumstances of her becoming ill, thus staying in the narrative a bit longer. Shortly thereafter, they change the topic. After about 5 min, the HCP returns to the subject, as manifested in the excerpt below (Table 6).

In Excerpt 5, the HCP returns to the topic of what caused the patient's sick leave, as was initiated in the previous excerpt (Excerpt 4, Table 5). Throughout the sequence, the HCP poses several questions directed both at the factual circumstances at work (Lines 1–2, 24, 26–27) and at the patient's understanding of what happened (Lines 18–19, 29). Here the HCP does not use open-ended questions, which are traditionally seen as the ideal practice for encouraging narrating.²⁹ Her questions are instead delivered in the form of assertions, which directs the patient's subsequent turn and prompts her to continue by accepting or declining the embedded suggestion. The first suggestion, in Line 1, is followed by a confirmation, in which the patient elaborates in more detail on how the stress at work affected her. Refraining from the open-question format produces another kind of relation to the narrative, one in which the previous content is challenged (Line 1–2 ‘but did you feel stressed’, Line 24 ‘but you didn't get any support’, Lines 26–27 ‘but did you ever talk to your employer’), until alignment in what happened is reached (Lines 34–39). In the patient's subsequent turn, the previously noticed uncertainty (Excerpt 4) towards what happened is no longer evident, as she either aligns (e.g., Line 20–23) or not (Line 28) with the HCP's suggestions. Although the HCP provides suggestions, it is the patient who, in her turns and in her own words, identifies the conditions that made her situation unbearable (Lines 20 and 22). Towards the end of the sequence, the HCP reformulates her understanding of what the patient has told her, acknowledging the story's credibility and validity, and by using the pronoun ‘you’ (in the sense of a general, impersonal pronoun), she normalizes what happened to the patient as something which is relatable and understandable also in a general sense. Throughout this excerpt, the story of what happened becomes clearer, challenging the uncertainty and personal failure-tendency visible in the patient's initial account.

3.7 The narrative as a means for joint reflections

Excerpt 6, below (Table 7), is taken from another conversation and presents a slightly different example of a co-constructed narrative also concerning a retrospective understanding of what happened before the sick leave. The excerpt begins in the middle of a longer narrative in which the patient reflects upon different aspects of his life and their potential connection to his recent illness and related struggles.

In Excerpt 6, the patient narrates the circumstances preceding his sick leave to the HCP. At the beginning of the sequence, he reflects on the topic of why he got a depression and when it started, while  also describing a process of change he is currently in the midst of. In Lines 12–16, he begins introducing present tense verbs, signalling the potential wrapping of this particular narrative. The hints on an exit are reinforced in Lines 18–20, with his summarizing statement that it feels good getting help sorting this out. In Lines 21–22 there is a brief moment of overlapping speech, but by switching to giving listener support (Line 23), the HCP lets the patient continue his turn, and he gives a clearer story exit in Line 25. Thereafter, a change in speakership occurs, and through the rest of the sequence, it is the HCP who delivers the extended turns and gives a tentative interpretation of what the patient has described happening. She explicitly positions her interpretation as tentative by referring to her being the second-hand source of experience (Line 32–33, ‘from what you've told me’), and continues on to use this interpretation as a base for supporting the solution the patient has found to problems in his current situation (Lines 34–42). Beginning in Line 46, by referring to the patient's process as ‘a journey’, and in Line 48–49, by suggesting that what he describes is a turning-point, she addresses aspects of the patient's overall illness narrative and contributes to his narrative understanding, that is, his meaning-making process. In Line 51, she signals that her reformulation is finished, and explicitly requests the patient's perspective on what she has suggested. He confirms her contribution and continues to elaborate on the topic (outside excerpt).

In contrast to a patient-driven narrative, this narrative is not told as a completed story with its morals already sorted out. Rather, both participants relate to the narrative with reflections on it. Co-constructed narratives have this participatory evolution of the story in common, although, as we have shown, the incentives can differ. In Excerpts 4 and 5 (Tables 5 and 6), the narrative of what caused the sick leave evokes emotional reactions and, initially, a general sweeping chain of events. Lingering on the topic, and returning to it later on in the conversation, shows an initiative from the HCP to get the patient to elaborate on the background story. In Excerpt 6 (Table 7), this elaboration is driven mainly by the patient's own reflective stance towards what happened, and the HCP participates in the reflective process with tentative interpretations.