Dynamic and distributed exchanges: an interview study of interprofessional communication in rehabilitation

Abstract Purpose Interprofessional communication (IPC) in rehabilitation is important for patient care yet it has been shown to be variable and challenging. Existing research does not address the complexity of IPC in this setting. Understanding the influence of contextual factors on IPC may guide improvements to increase the effectiveness of communication within interprofessional teams. Methods From July 2020 to February 2021 semi-structured interviews were conducted with 24 healthcare professionals across Australia and New Zealand. Cultural Historical Activity Theory provided a guiding theoretical and analytical framework for this qualitative study. Results Participants described engaging in IPC through evolving interactions, piecing together information that underpinned patient care. Meetings occurred frequently, however communication extended well beyond formalised interactions, often requiring individuals to balance clinical workload with communication tasks. IPC reportedly relied on communication tools, however navigating information from multiple sources was demanding. Conclusions Our results indicate that IPC contributes significantly to the workload of healthcare professionals in rehabilitation. IPC was integral in the provision of cohesive patient care, however it proved time consuming, effortful and at times frustrating and potentially erroneous. Our findings promote the need for rigorous examination of communication practices to ensure they are meeting the needs of an increasingly interprofessional workforce. IMPLICATIONS FOR REHABILITATION Healthcare professionals should recognise that time spent communicating within their team is a legitimate and important part of patient care. Rehabilitation teams should consider how they allocate resources for communication tasks. Teams should reconsider how they can use communication more effectively to save time by reducing repetition and errors.


Introduction
Interprofessional communication (IPC) is a critical component in advancing healthcare delivery and improving patient health outcomes [1-4]. However, in inpatient rehabilitation it can be variable and challenging due to the complicated presentation of typical patients, their length of stay, and the number and diversity of healthcare professionals (HCPs) working together [4,5]. This setting provides intense, interprofessional rehabilitation intervention for patients whose current functional status is preventing them from residing safely in their usual accommodation [6]. As such, a large amount of information is exchanged between HCPs of different disciplines and between HCPs, patients, and their family members [4,7,8]. When this is goal driven communication relating to patient care it is defined as IPC [9,10]. Research on IPC in healthcare, including rehabilitation settings, shows that it is at seek to examine such contextual influences as we address the research questions: "How do members of a rehabilitation team engage in IPC?" and "What factors influence the perceived effectiveness of IPC?" Understanding the influence of context in relation to IPC is essential to adequately address existing barriers and to improve the communication in interprofessional teams.

Methodological framing
In this study, we have taken an interpretivist approach, positioned within a socio-cultural view of the clinical context, in an effort to deepen understanding and gain insights that might be used to improve practice.
Aligning with our interpretivist stance, we used Cultural Historical Activity Theory (CHAT) as a theoretical lens to address the gap in the current conceptualisation of IPC in healthcare [3]. CHAT is a theory developed to study complex social interactions in which people's activities impact reality as part of ongoing culturally and historically situated, materially and socially mediated processes [28]. To study activities, CHAT-informed research refuses single-cause explanations for activities, instead attending to the "collective nature of human activities" by examining activity systems [29][30][31][32]. Activity systems allow deep examination of the factors which shape human activity, including tools, other individuals (subjects), communities, divisions of labour, social rules, and the goal(s) or objective(s) to be achieved [19,30] (see Figure 1). Activity systems also highlight tensions and contradictions that may exist between these factors. These features mean CHAT has particular relevance for studying complex interactions, such as IPC within healthcare teams [3, 30,33].

Researcher positioning
The methodological framing of this study reflects our research experiences and worldviews. Within the team we have a diverse mix of clinical and research expertise. JP has extensive experience working as a physiotherapist in inpatient rehabilitation which afforded her "insider" knowledge and allowed her to understand and use the language of healthcare practices during participant interviews and data analysis [34]. CD and EM both have a background in physiotherapy (although not primarily in rehabilitation) and are experienced medical education researchers while RWK is a healthcare communication researcher with a background in linguistics. The experiences of the research team and the CHAT perspective influenced how we saw the types of texts produced and engaged in. This guided data collection and our interpretation of the data.

Participants and sampling
Twenty-four qualified HCPs took part in this study, most participants were female (75%, n ¼ 18) and 58% worked full time (n ¼ 14). Purposive sampling targeted participants with particular characteristics [35]. These included professions (nine nurses, 11 allied health and four medical professionals were interviewed, see Table 1 for details) and inpatient rehabilitation settings (a variety of patient cohorts, private and public hospitals in both metropolitan and regional locations). Twenty participants were from Figure 1. A visual depiction of an activity system (adapted from Engestr€ om [30]). This figure shows how a subject's (being an individual's or group's) work towards the achievement of a specific objective is tied to and influenced by the tools that are available to complete the task, the rules and norms of their social communities and their conceptions of division of labour.
Workplace settings and locations have been removed to protect participant anonymity.
Australia and four were from New Zealand, one of whom drew upon workplace experience in Canada. All had at least six months experience working in inpatient rehabilitation, and many had worked in this setting for over 10 years (58%, n ¼ 14). After ethical approval, participants were approached via email through the research team's professional network. Individuals were sent a brief email introducing the project and further recruitment information was only sent on request. This allowed individuals adequate time and ability to freely consider whether they wanted to participate. As none of the research team are practicing health professionals there was no professional pressure for individuals to agree to the study.
Sampling in this study was guided by the principle of information power [36]. This principle argues that a broad aim and crosscase (between participant) analysis, which is characteristic of this study, indicates the need for a larger sample size to maintain information power. Conversely, a high sample specificity and a strong dialogue within all interviews indicates a greater information power and therefore a smaller sample needed. Based on these considerations, a provisional number of 20-30 participants (medical n ¼ 5-8, nursing n ¼ 5-8, and allied health, which consists of many different professional groups, n ¼ 10-14) was the initial sample size for our research project. Appraisal of information power was repeated throughout the process, and the total number of allied health and medical interviews conducted (n ¼ 11 and n ¼ 4 respectively) reflect that data collected from these participant cohorts had adequate information power for a rich analysis. The relatively large number of nurses included in the study (n ¼ 9) reflects that the information power from this cohort was lower, requiring a larger sample size to examine nurses' experiences to address our research questions.

Data collection
The primary researcher (JP) conducted all interviews between July 2020 and February 2021 online (n ¼ 17) and over the phone (n ¼ 7). These methods enabled collection of rich data through synchronous interactions despite restrictions in place due to the global COVID-19 pandemic preventing face-to-face interviews [37]. Semi-structured interviews were used, with an interview guide that included initial questions and probes (Supplementary materials Appendix 1) informed by IPC processes identified in the scoping review authored by this research team [4]. This schedule provided a framework of questions that allowed the interviews to be conversational while pursuing complex, nuanced descriptions of IPC [38,39]. The interviews incorporated the use of a visual aid, a discourse map derived from the scoping review findings, which was used to trigger verbal responses from participants (see Figure  2). This process involved the researcher editing and adding to the existing map to capture participants' responses, thus co-creating another source of data [40].
Participants were asked to share demographic information, describe IPC exchanges they routinely engaged in (prompted by the discourse map) and to share experiences of IPC to explore their interpretations of what constitutes effective communication (i.e., "In your role, what interactions are most important to your daily practice and why?"). Participants were invited to elaborate on what interactions they find most useful and to give examples of recent interactions. Notes were made on the discourse map during the interview to reflect participant responses and probing questions were used to clarify details throughout. After each interview, the primary researcher completed a "reflexivity memo" documenting immediate impressions. All interviews were audiorecorded, transcribed verbatim and were anonymised. Discourse maps were downloaded, deidentified, and then linked to deidentified participant transcripts. The written transcripts, discourse maps and reflexive memos became the final data set.

Data analysis
Data analysis was guided by template analysis, and we concurrently collected and analysed data in an iterative process. The primary researcher (JP) recorded notes as part of the sense-making process during analysis and these were discussed with the rest of the research team throughout data collection. JP led the analysis and had frequent meetings with the research team to discuss data analysis. Template analysis, a form of thematic analysis, was chosen to develop a structured coding template. This analysis method, as described by Brooks et al. [41], follows six steps: (1) familiarisation, (2) preliminary coding, (3) organisation of emerging themes, (4) defining an initial coding template, (5), applying initial template and modifying as needed and (6) finalising the template and applying it to the full data set. Familiarisation (Step 1) involved JP listening to and analysing the first five interview transcripts. This was followed by preliminary coding of this data subset (Step 2), which was discussed with the research team (RWK, EM, & CD). Identified themes were then organised (Step 3), and at this stage it was determined that CHAT could be used to further examine the data. We then reorganised the themes (repeating Step 3) into an initial coding template using CHAT constructs as overarching themes (the six factors represented in Figure 1). The data was then iteratively coded within these themes and sub-themes were developed from prominent concepts within the data. Iterative rounds of coding were completed to identify how the data related to CHAT and themes were modified, redefined, and added. For example, the themes of "community" and "outcomes" were consolidated within other activity theory themes as there were few data coded to these themes that did not also fit elsewhere in the template. An additional theme "communication actions" was developed to represent the types of communication described by participants, and "evaluations" was added to capture participants' positive and negative evaluative comments. This resulted in an initial template which was applied to six transcripts (Step 4). These were coded independently and then discussed in research pairs (JP & RWK, JP & EM, JP & CD) which led to further modification and refinement of themes (Step 5). Further team discussion led to consensus on a final coding template (Step 6) which was uploaded to NVivo and applied to the full data set (see Supplementary materials Appendix 2 for the full coding template). The final coding template was then used to guide interpretation [42], whereby data were interrogated to explore patterns and associations between sub-themes.
Throughout the analysis process we were struck by the way participants described connections between IPC processes and the iterative exchanges of information, which we had coded as "stepping-stones" in the final coding template. There was a strong emergence of "distributed" IPC interactions; interactions that often occurred in different locations or at different times and that built upon each other (in contrast to discreet communication episodes as represented in previous rehabilitation literature) we reached to Engestr€ om's concept of "knotworking" as a way to help explain our findings. Given this strong resonance, the "knotworking" theory helped inform our late-stage analysis.
Knotworking, a fairly recent development within CHAT, describes collaborative work as situations involving constantly changing combinations of individuals distributed over time and space [30]. To study social interactions resulting from continually and dynamically changing conditions, Engestr€ om et al. [32] developed this concept and later defined it as "a rapidly pulsing, distributed and partially improvised orchestration of collaborative performance between otherwise loosely connected actors and activity systems" [30,p. 86]. During a collaborative interaction, individuals bring a unique perspective (their own activity system) which can be thought of a distinctive "thread" of activity. When these threads tie together, this is a "knot." The knot can continually change and may involve the threads of different individuals at different times and as such there is no centre that fixes the collaborative efforts of those involved. The knot cannot be reduced to a particular individual or organisational centre of control [30]. Instead, the locus of initiative changes from moment to moment. This additional theoretical perspective of "knotworking" informed our final analysis for RQ1, providing a lens to examine the relationships between sub-themes. These relationships were not linear, so rather than refining sub-themes into a set of simpler categories, we developed three key categories which described the most pertinent research findings. Data addressing RQ2 were further interrogated to explore patterns and associations relating to participants' perceptions of effectiveness. The resulting data structure is illustrated in Supplementary materials Appendix 3.

Results
This study sought to examine how HCPs engaged in IPC and to identify factors influencing effectiveness. In this section the results are presented within the categories identified through analysis. As participants discussed their experiences of IPC it became evident that HCPs spent a considerable amount of time engaging in IPC in the form of scheduled meetings, ad hoc interactions, and written communication. Participants explained that IPC was used to navigate care needs as they evolved through patient admissions, often elaborating on experiences from current and previous rehabilitation roles and from their observations about others involved in IPC, including patients and care givers.
Overall, HCPs engaged in IPC through iterative interactions, akin to Engestr€ om's [30] description of "knotworking." IPC often involved a multitude of meetings, however much communication also occurred "behind the scenes" which required HCPs to balance their clinical workload with additional unscheduled communication tasks. Engagement also relied on communication tools which are the physical objects or modalities used as part of a communication process [43]. HCPs described using a variety of tools to straddle temporal and spatial barriers to communication. In the following sections results pertaining to RQ1 are outlined in three key categories: (1) information was pieced together iteratively, (2) meeting load, and (3) navigating communication tools. Results relating to RQ2 are incorporated within the second and third categories as participants explained that effectiveness was influenced by the perceived workload associated with IPC and the ability of communication tools to facilitate sharing of information amongst team members. Evaluations were nuanced and varied between participant accounts and even within individual participant's data. Both positive and negative evaluations were described within each category and have been included to represent the data.

Information is pieced together iteratively
Participants spoke of IPC occurring during interactions that were nuanced and interconnected; it did not occur during isolated, discrete episodes. A nurse participant explained how she obtained relevant information using a combination of verbal and written sources: IPC was said to occur over lengthy trajectories of time, through formal, informal, synchronous, and asynchronous processes, and over distributed spaces (e.g., allied health treatment areas, patient bedrooms and non-clinical spaces). The following example illustrates how "snippets" of information were pieced together through interactions with multiple individuals: [The patient's friend was visiting] and let me know about the discharge plan to a residential facility. So, I then took the opportunity to confirm all that [with the patient's family] and then double check with the [patient] this morning to see whether she'd remembered the conversation … So, you know, little snippets of information you need to follow through with that you probably miss otherwise, if you weren't there at the time of that initial discussion … And I then relayed that via email to the social worker who I know is very involved trying to get this lady to the right place at the right time. So, she was appreciative … one little thing, you know, is very relevant information to another person in the team. (RN5) Our participant accounts affirm that IPC occurred through interactions that ranged in formality which were influenced by rules associated with each interaction. Some interactions were relatively informal with few rules explicitly designed to structure communication: [Staff] may see the family members and say "Are you right? Oh, you want to have a catch up? We'll catch up. How about tomorrow?" And send a quick group email. It could be a very ad hoc, informal sort of thing. (NUM1) Other interactions involved the use of prefigured rules and processes. This is outlined in the following example about weekly team meetings: The consultant would normally chair the meeting … the goals are recorded in the electronic medical record [EMR] system, and they're read out … Certain people have certain roles. (PT2) IPC was described as a dynamic process, acknowledging the contribution of both ad hoc and formal processes to the team's overall understanding of a problem or situation: Because [the patients are] so complex, we need all kinds of communication … Informal and formal. We use all of them daily together. We can't miss one of them … it's a combination. We can't miss any. (RN2) Participant accounts showed that IPC was a dynamic process, comprising of successive and parallel task-oriented combinations of people and communication tools, just as Engestr€ om [30] described in the "knotworking" concept.

Meeting load
Participants described heavy scheduling of meetings and that the workload associated with meetings often extended well beyond the time allocated to each interaction. Meetings included daily handovers (or huddles), discipline specific and team planning meetings, weekly team meetings (also known as case conferences) and ward rounds in addition to other regular meetings (family meetings, complex discharge planning meetings, falls reviews, and non-clinical meetings). The need for so many meetings was attributed to the extended length of stay in rehabilitation (often lasting weeks to months), and frequent meetings were also used to iteratively share information which influenced patient care provision. Hence, the combination of people and contents of meetings changed often: The first meeting of the morning would be typically OT [occupational therapy], PT [physiotherapy], and rehab assistant … setting the plan for the day. And then the speech pathologist and social worker would … join in when you then converged … on the actual unit … Then there'd be lots of kind of discussion around there. And then bullet rounds [daily handover] would start with the rest of the team and nurses. (PT4) Attendance in meetings depended on professional roles and patient need. This participant reported how her team engaged in additional meetings to address the complex problems within their patient cohort: When you have a complex patient, we'll flag them at the Monday, Tuesday, or a Thursday [team meeting] … we're like, "there's a lot going on there," and we probably can't cover them in the five or ten minutes we have in the Monday or Tuesday meeting. So, we arrange to have a meeting … just the clinicians involved … So, for some of the complex patients who have lots of involvement from lots of different clinicians, we'll set them for every week … . sometimes we sort of figure out what we're doing, and then they drop off … . I think I had three in the four days I was there this week. So, rehab is very meeting heavy. (SW1) This participant went on to state that her team had a "ridiculous amount of meetings"; however, she also reflected that they "help ensure we're always pretty much on the same page" (SW1). This was echoed by other participants, although when the purpose of meetings was unclear it was suggested they were less useful, and that "it can cause a bit of tension... [people felt] it was a waste of time" (SP1). Others explained that it was challenging to balance attending meetings with their clinical tasks: Although not everyone attended weekly team meetings, all participants reported they were a consistent feature of rehabilitation IPC. The following example outlines that these were a valued component of patient care: This sentiment was reflected across the data as participants described diverse team meeting attendance, processes, and procedures. Meetings were usually organised around the availability of the medical consultant and in circumstances when a ward had more than one consultant, multiple meetings were scheduled per week. Unsurprisingly, consultants were the most consistently represented profession in addition to nurse unit managers (NUMs) or ANUMs. The attendance of allied health and registered nurses fluctuated and often depended on staffing: I definitely am involved family meetings … being a smaller discipline and me having to go to every family meeting takes a lot of time. So, I would be going to the family meetings where there's significant dietetics involvement … [If I don't attend] I usually … email one of the other key allied health members and I give some dot points summaries in lay persons terms about the patient's progress. (D1) It emerged that often a representative attended meetings on behalf of a whole professional group, or a written handover was provided prior to the meeting in order to provide discipline specific information. The following example highlights the additional communicative work that often helped scaffold weekly team meetings. In this case it involved intraprofessional interactions (within the same profession) which contributed to the sharing of interprofessional information: The consultant goes [to the meeting] and then a representative from each other discipline … So, if you're the OT that attends it's your responsibility to get a handover from your OT colleagues and hand back anything that the team raised that's relevant to that OT … We have a handover document that we made … Before the meeting, all the different OTs will go in update their patient's details and goals and handovers [on the document]. And that's also the document that the OT who's in the meeting types back the feedback into. (OT2) The interconnection between intraprofessional and interprofessional interactions was evident throughout the data. Although often reported as necessary, participants outlined the additional workload of such interactions and at times it was also a source of confusion and error: It can be frustrating. Because, when you're doing your handover [in preparation for the team meeting], you're writing what you … think is going to be talked about [in that meeting]. Say you have covered results of a cognitive assessment that you've done but then the team asked, "Do they have stairs?" Or something like that … So, the team finds that frustrating as well. Because there's an OT [present] who can't tell them anything about the environment because you haven't written it in … It does break down at times. (OT2) In this case, it appears that conflicting clinical priorities within the team resulted in compromised communication during the meeting. Expanding on this comment, this participant went on to define additional tasks that were required to resolve the communication breakdown after the meeting, further adding to her workload.

Navigating communication tools
HCPs indicated that communication tools were a routine part of IPC, however they described a tension between their utility and the effort required to navigate multiple tools while adhering to established practices and team preferences. Tools included technologies such as phones, computers (used for email or to complete electronic medical records), and electronic pagers as well as analogue tools such as hardcopy medical records, forms, and charts. They were relied upon for formal documentation and facilitated ad hoc interactions and correspondence between team members and with patients and their families. Tools seemed to be particularly useful to enable teams to engage in IPC across different locations and when work schedules clashed: This participant reflected that the co-location of staff influenced the utilisation of communication tools, whereas the following example indicates how tools were used to access information remotely: I'll often ring up at … eight o'clock at night and say, "Look, I'm just sort of touching base, how is everybody?" You know that goes on a bit … Nursing staff will ring me, you know, it's "I'm a bit worried about this patient's behaviour." (MC2) Many participants revealed their preference for face-to-face conversations, however, communication tools facilitated IPC with patients, families, and other HCPs when this wasn't available. The following example shows how tools were used as part of a workaround strategy to engage with other HCPs when clinical roles impacted their ability to communicate in person: On a daily basis I would have a lot of informal communication with nursing and medical … but with the [allied health] professions … often I can't get them at a [convenient] time … So I would usually page them and say, "can you please call me when free regarding this patient" … and an extension to call on. (D1) Often, as above, written and verbal communication were interwoven as HCPs worked within the constraints of their specific workplace to access or share information. However, the majority of tools conveyed written information which participants described as critically important to their work. When written information was accessible and up to date it was highly valued, and participants described that accurate documentation provided an opportunity to share information despite conflicting work schedules. One participant summed this up: The formal written communication is so, so important because … . when you have no one to ask questions and when no one knows the answer … . We will always go back and have a look at the documentation. (RN4) However, others indicated that navigating written communication was often challenging due to inaccurate, repetitive and discipline-specific jargon which impeded accessibility to information, as explained here: As in this case, participants explained that they often spent considerable time and effort navigating different communication tools to source accurate information. The following example reveals how conflicting information resulted in frustration, significant cognitive load, and additional clinical work for the nurse participant: I was nearly pulling my hair out because … Everyone's got [a whiteboard] at the bedside … their therapy timetable … And we also have a printed timetable in the handover room and pinned up on the notice board … [On the bedside timetable] a lady … had speech at 10:30 and physio at 11:30. That wasn't on the [other] list … And I thought, I'll err on the side of caution, and I'll get this lady organised. So, I did do that. And her bedside timetable … was incorrect." (RN3) Tools such as printed timetables, goals sheets, and whiteboards were mentioned frequently, and it was suggested their function was mainly to supplement verbal interactions with patients and families. However, patient engagement with these tools appeared varied. Some participants questioned the functionality of communication tools for patients, such as when "the patient's timetable … that's in small font often behind the patient's head … . No one can read it" (SP1). In contrast, it appeared that at times careful consideration had been given to incorporate patient and family input into IPC, helping to align expectations: We have paperwork that you go through with the patient … to elicit questions that might help to ascertain their goals … . We also have a questionnaire that we provide … for the family to fill out to help support in goal setting as well … what they expect their loved one to be like when they leave us too. It also asks questions like "how do you feel you will achieve your goals? What needs to happen?" … . Our goal setting is very much what the patient has stated as their goal. And it's in their words … we don't re-word it or use technical terms … It's very much what the patient has stated. (PT4) Engagement with patients' families was severely impacted by strict COVID-19 restrictions which limited their ability to visit loved ones. Participants indicated an increased reliance on communication tools as they engaged in significant workaround strategies to overcome challenging circumstances. Communication with patients' families was said to become heavily dependent on phone calls, emails, and video or teleconferencing. The following example shows strategies a participant used to engage with families despite COVID-19 restrictions: Families weren't allowed to visit, you know, you got people who are … severely brain injured, and they weren't allowed to visit. So … I would get the family member's phone number and take some video of [the patient] standing up or walking to the first time and then, you know, just share those videos. (PT5) When communication solely involved professionals within the team, pager messages, texts, and phone calls were seen as useful when prompt sharing of information was a priority. Email was heavily utilised and was described as a useful way to disseminate information; however, timely information exchange through email was challenging and many participants expressed being "flooded" with emails. This participant vented that emails were often used as a way of "avoiding a conversation that may not be an easy conversation to have" (SP1), and went on to give this example: A junior doctor, he was fairly new to the ward, sent an email out about five minutes before allied health were due to leave. Saying "this patient has expressed suicidal intent and has been specific about how he's going to do it. I just thought I'd let you know, maybe we should discuss discharge," and we were like, "oh my god, are you serious?" So, I think there can be inappropriate use of emails. (SP1).
As seen here, participants identified that communication tools were problematic when they were poorly used or when a tool "just doesn't work the way it should" (NUM1). Challenges were even apparent when participants described using medical records which were the most frequently mentioned communication tool across the data. Despite their established use, at times records appeared hazardous; problems related to the record itself (e.g., poor functionality of electronic medical records) and to the way people completed documentation in the record: Some things are documented eight or nine times [by different professions] … I don't know why we have to repeat all that … in the [medical record] notes, then on our physio assessment form, and then in our weekly handover. [We should just write] "injuries noted as per medical assessment." (PT5) The following example details the workaround strategies a participant engaged in to overcome limitations of a new communication tool implemented in her workplace: There's a [new] referral system where the doctors are meant to formally refer on [EMR] if a patient needs to be seen [by dietetics] … Often, doctors will re-refer a patient to me that I'm already seeing so I don't think they're reading my notes at all … Referrals can happen, by pager, in person … To make sure I'm diligent, I'm screening the list. And I'm listening in hand over to new referrals … My biggest concern is that I will get referred a patient once they're already malnourished. (D1) It was common for participants to describe engaging in individual or team workaround strategies when they lacked confidence in written information. It appeared HCPs did so to ensure patient needs were met and, in some cases, to avoid serious medical errors: Currently, we're having a lot of errors [with electronic medical records], a lot of medication errors … drugs dropping off the system. The other day one of the staff members came along, just said "This patient said her meds are wrong." … And we looked at the previous day's medication chart … suddenly these medications were appearing on that chart. If you go to today's chart … they'd dropped off completely. (NUM1) Participants described coordinating care activities through documentation and shared communication tools, use of which was often improvised, or were contingent on a team or individual clinician's preferences. These tools were described as an integral component of IPC, particularly due to the asynchronous function they afforded. However, it appears that individuals were often juggling the use of many communication tools, an additional burden that required participants to interpret potentially conflicting information.

Discussion
In our study of IPC in inpatient rehabilitation settings, HCPs engaged in IPC by integrating information iteratively through a multitude of meetings and using a variety of communication tools. The way participants described information being "pieced together" strongly reflected patterns described as "knotworking" by Engestr€ om [30]. Although all teams relied on formal meetings, spontaneous, emergent interactions, documentation, and a variety of communication tools as part of IPC, there were significant differences in how participants perceived their effectiveness. Effective IPC was seen to be achieved when processes helped team members understand each other quickly and coordinate patient care efforts. This reflects what Engestr€ om [30] describes as successful "knotworking." Conversely, IPC was reportedly less effective when participants had a heavy meeting load and needed to engage in time consuming workaround strategies to navigate communication tools. Here we see a point of departure between our data and "knotworking." Although our results showed that all IPC involved similar iterative processes, the trajectory varied considerably and often relied heavily on the dedication and commitment of teams or individuals for successful IPC.
Our findings demonstrate that rehabilitation teams pieced together information by engaging in IPC. This contrasts with much of the literature on communication in rehabilitation that infers that the "important work of communication" plays out in scheduled meetings where there is representation of the different disciplines working with the patient [4,5,44]. Within the rehabilitation literature, weekly team meetings are symbolised as the event for the exchange of information and decision making [44]. However, the findings from this study suggest that team meetings are only one component of IPC, bringing into question the nature of decision making in this setting. Engestr€ om [30] argues that "knotworking," as a form of collaborative expertise, is essentially a form of decision-making. All our participants gave examples of communication interactions that involved decision making outside the context of formal team meetings. The implication is decision making through IPC reflects an "ecological" vision of decision making [45] which is an important shift in our understanding of how organisational decisions are made in this context. The idea of distributed communication and decision making rejects traditional views that decisions are made at a given point of time in a particular place, involving specific individuals [45]. Such views underestimate the systemic character of decision-making in organisations. As Mehan [46] articulated, often decisions made in formal meetings are the culmination of a lengthy process and this is clearly reflected in our data reflecting the rehabilitation context.
There are several key issues that were identified regarding the interaction of CHAT activity system factors (Figure 1) that could be explored further to support a nuanced understanding of IPC. However, the within the scope of this paper the most significant finding related to communication tools and the pivotal role they played in the effectiveness of IPC. Our participants highlighted that communication tools were often the lynch pin holding distributed IPC together. When the communication tools available reduced the effort and cognitive load associated with IPC, HCPs were willing to trust shared information and were able to engage in "knotworking." It has been shown that the combination of activity systems involved in healthcare activities are inherently unstable, meaning that communication tools used to coordinate actions are of particular importance [3]. In their paper proposing the use of CHAT in the investigation of IPC and medical errors, Varpio et al. highlighted the relevance of analysing communication tools used to coordinate medical care [3]. They warn that confusion or errors generated when multiple communication tools are shared by members of interprofessional teams can easily become sources of latent medical errors [3]. Our findings provided similar cautionary examples. Errors may be enabled by the significant burden on individuals to synthesise information from so many sources as well as the risk for misses and ruptures in "knotworking" posed by having to navigate multiple tools.
Our results suggest that healthcare organisations should examine the function and utility of communication tools to reduce the communication workload for rehabilitation teams and minimise communicative errors. This research shows that CHAT can provide a useful framework for such evaluation as it calls attention to the complex interdependencies of a communication activity's components including how specific tools are used [3,30]. In order to construct solutions that support the needs of all team members it is relevant to consider the utility of each communication tool and how team members incorporate them to piece together information [3]. Another way of thinking about this implication of our work is to consider the ability of communication tools to function as "boundary objects" which link the diverse interests that exist within a team [47]. Communication tools can play an essential role in crossing boundaries that exist between professions and between HCPs, patients, and families [48]. However, tools can fail as boundary objects when they do not fully capture multiple perspectives and are not easily understood by all individuals involved, such as when patients were provided written material using complex language and medical terminology [48]. Our results indicate that concerns and interpretations of all users are important when considering the design and implementation of communication tools, especially those being used by patients and their families. It has been argued that boundary objects can be perceived or used differently over time, at one time enabling communication and collaboration, whereas at other times losing their boundary crossing function [48][49][50]. Findings from this study highlight that regular evaluation of communication tools could be useful to ensure that tools continue to act as boundary objects as teams evolve over time.
What recommendations can we make to facilitate IPC and improve its effectiveness? Changing individuals' communication may help but is unlikely to be sufficient. We have shown that HCPs spend an extensive amount of time communicating, so it would be useful for teams to collectively reconsider how they use IPC to increase quality as well as efficiency. Developing local communication guidelines is one way to promote more effective IPC. For example, deciding as a team to use jargon-free language or that a patient's past medical history is only documented once by an allocated clinician (i.e., the admitting doctor) are changes that have potential to save time by reducing repetition and errors. Another example is to have a brief introduction at the start of weekly team meetings to reinforce the goal of the meeting and any communication guidelines that might be in place e.g., if there is insufficient time for extensive clinical discussions, staff should be reminded of this at the start of the meeting. Regular re-evaluation of processes and education of new staff members are also important to align expectations within a team. Small changes to local practices can help develop a positive communication culture that further promotes effective IPC [4].
The timing of our study gave interesting insights that showed that evaluation of communication practices and tools during periods of drastic organisational change could prove particularly important. First, many participants reflected on how the implementation of new technologies had unexpected influences on IPC. Second, our data showed the adoption of emerging practices as healthcare organisations navigated COVID-19 restrictions. These entailed an increased reliance on communication tools and an expansion of decision making in the spatial and temporal dimensions. Resonating with our findings, others have identified that the distributed features of decision-making are highlighted in periods of intense change [30,51,52]. In such periods, routines are called into question and new procedures are articulated [30,52]. Although the current climate poses challenges for healthcare organisations, from a communication perspective it can be seen as an opportunity for change. It has been noted that from an organisational perspective, this is an extraordinary moment in which at some level, the metaphorical "decks have been cleared" [53]. The findings from this study provide further support for the need to pause and critically consider what has been learned in the COVID-19 pandemic; what organisational practices should be reinstated and what adapted or new practices should be maintained.
Determining whether communication practices are meeting the needs of an increasingly interprofessional healthcare workforce is imperative. Healthcare teams and organisations must recognise that distributed communication practices are an integral part of interprofessional teamwork and that adapting to changing demands may require a reconceptualisation of communication practices.

Implications for research
We found that HCPs pieced together information iteratively, through evolving and intricate encounters which often occurred outside formal interactions. Our findings resonate with those of Ellingson [54] who used Goffman's concept of front and backstage to explore communication practices of a geriatric oncology team. Ellingson's research revealed the critical importance of "backstage" communication that occurred outside of formal team meetings [54]. Expanding on this work, our data demonstrate the considerable workload associated with the "backstage" episodes that contribute to IPC in rehabilitation. These findings reinforce the need to closely examine such communication episodes to better understand how they contribute to shared knowledge and decision making in rehabilitation teams. While hard to quantify, research that legitimises these "backstage" interactions could help teams advocate for adequate time and resourcing to engage in essential, but difficult to see, patient care activities.
Another important insight from this research is that IPC should be viewed as a progression of integrated and dynamic interactions. As such, researchers should consider empirical analysis of the entire, networked "path" of communication to understand how it contributes towards decision making in an interprofessional team. This idea is closely related to the concept of "trajectory" coined by Anselm Strauss [55,56]. In their study of distributed actions in healthcare, Strauss and colleagues used a patient's entire illness trajectory as the empirical unit of analysis [55,56]. Such research is needed to examine distributed IPC, including interactions that may influence decision making without specific intention. Through the lens of CHAT, all activities are saturated with decision-making actions, including those that are relatively invisible and involve little conscious articulation [30]. Research examining the trajectory of communication within interprofessional teams is needed to gain understanding of how small, seemingly inconsequential interactions occurring over time may influence the direction of patient care.
In this study we have explored the narrative accounts provided by participants to examine the interpretive layer of causality in relation to our research questions rather than analysing the relationship between IPC and a particular result or patient outcome [57]. Future research could expand on these findings by incorporating observational data collection to gain insight into actual practice which would facilitate the development of rich, deep, and holistic understanding of IPC.

Strengths and limitations
We have involved participants from across both Australia and New Zealand which we think is a considerable strength of our study. Investigations focused more narrowly into differences between individual rehabilitation teams, or more broadly in other countries are possible avenues for future research. Interviews with patients and family members would provide important, supplementary perspectives on themes identified in this study. In our research design, we have made choices in our approach and the use of the CHAT theoretical framework as a deductive overarching framework. We acknowledge that this has sensitised us to themes within the data that aligned with factors foregrounded by this particular theory. Rather than being a limitation, we believe our rigorous coding process and transparency in reporting contributes to the robustness of the study. In future, other researchers could further enhance our understanding of IPC in rehabilitation by applying other social theories that illuminate different facets of communication.

Conclusion
In order to be impactful, research into IPC needs to reflect the inherent complexity of communication; acknowledging its distributed nature which involves many individuals and often relies on a multitude of communication tools. This paper used the richness of HCPs' experiences and perspectives to illustrate that teams engaged in intricate and evolving IPC as they pieced together information to support patients through their rehabilitation inpatient stay. Employing CHAT, with its concept of "knotworking," allowed us to identify and explain that the workload associated with IPC and the tools used to enable information sharing within complex teams influenced the perceived effectiveness of IPC in rehabilitation teams. Our results suggest important expansions to current ways of conceptualising IPC that represent the intricacy of everyday communication in healthcare teams.