Specific studies on the prevalence of multimorbidity in New Zealand have not been undertaken. However, New Zealand is likely to have comparable numbers of those with multimorbidity to similar countries, such as Australia (one-third of the population), as the rates of long-term conditions are similar [26, 27]. In New Zealand, multimorbidity is likely to have a greater impact on Māori (New Zealand’s indigenous people), who have higher rates of chronic illness and overall lower life expectancy [28].

New Zealand provides no-charge secondary and tertiary medical specialist and hospital-level care, but there is a part-charge for patients to obtain first-contact general practice services (GPs and practice nurses) through a mixed capitation/fee-for-service funding model. Professionals, such as counsellors, medical specialists, midwives, physiotherapists, and social workers, work in either private or public health and social care agencies, with private agencies charging for service [29].

In the 1990s, the New Zealand government introduced lead maternity carers (LMCs) to provide funded maternity services. LMCs are either midwives, obstetricians, or GP obstetricians (there are now very few of the latter), and are chosen by the pregnant women to lead their maternity care. Some women attend GPs for confirmation of pregnancy and early antenatal care, but most approach an LMC from the start of their pregnancy. It has been argued that this form of maternity care isolates the LMC and disrupts the usual continuity of care provided by general practices [30].

New Zealand general practice and public hospital staff use different electronic health record (EHR) platforms. These platforms are used to record consultations, other interactions with and about patients, and are a repository for scanned letters, laboratory, radiological, and other test results. Some general practice and public hospital staff have limited access to each other’s patient EHRs. Electronic patient portals (patient e-portals; providing secure online access to personal health information) have recently been introduced in general practice settings [29].

Participants included two GPs, four patients, and three nominated professionals from different agencies; nine participants in total. The two GPs (GP1 and GP2), both experienced part-time female GPs (a typical profile for a New Zealand GP) [32] from two diverse practices (in terms of location, patient demographics, and business models), were approached, and both agreed to participate.

The two GPs were each asked to identify two patients with multimorbidity. The definition of multimorbidity, as defined by Le Reste et al. [3], informed the inclusion criteria of the study: “at least two ongoing chronic conditions requiring frequent care from their GP and at least one other health or social service agency, and be able to keep a diary for 12–13 weeks”. Each GP requested agreement from the patients to be contacted by a researcher. A research nurse then contacted the patients to explain the study and obtain their consent.

The GPs used purposeful sampling of patients to identify information-rich cases [33]. All four patients agreed to participate and complete a diary of health contacts over 12–13 weeks; no patient declined to participate. The patients were asked to nominate a professional from another agency with whom they had frequent contact during the diary collection period to be part of the study. All of those nominated were healthcare professionals and all agreed to participate. One patient did not have sufficient contact with professionals in other agencies during the data-collection period to nominate another participant. Informed consent was obtained from all participants.

Data for each patient were collected from four key sources, as outlined below (see Figure 1 for the sequence of data collection, analysis and synthesis).

1. Patient health encounter diary (indirect observation)

The patients completed a 12–13-week structured health encounter diary [34], which was chosen as an effective indirect observational method [35], as direct observation was thought to be unfeasible and unacceptable to patients over a prolonged period of time [36]. Methods of indirect observation, such as diaries and similar daily records, are considered comparable to those of direct observation [37] (the visual observation of individual participants in natural settings [38]).

The diary developed by the research team was piloted with two test patients and refined. It comprised a template of two distinct sections. Part 1 was a daily assessment of general health recorded on a 5-point scale, and a question about whether or not there had been any health contact made. Part 2 was an opportunity for guided free text entries of any health contact that occurred either face-to-face, by phone, text, letter, patient e-portal or email. Patients were also asked to comment on whether or not they thought the health or social carers might have communicated with each other. The ethnographically trained research nurse, who had extensive experience in collecting interview and observational data in primary care settings, kept in regular contact with the patients during the diary-keeping period, visiting in the initial weeks and then following up with phone calls.

2. Diary-informed interviews

Following analysis of the patients’ health encounter diaries, the research team developed an unstructured interview guide [39] to conduct individual face-to-face audio-recorded interviews with the patients and GPs. In contrast to semi-structured interviews, where interviews are often the sole data source and are based on a set of largely fixed questions, the unstructured interviews with the research nurse took the form of ‘guided conversations’ [31, 39]. These conversations centred on the completed diary, which was reviewed during the conversation. By the time the diary was completed, the research nurse was well known to each patient and had maintained field notes of multiple conversations during the diary-collection period. Drawing on information from her previous conversations, she was readily able to query the content of the diary with the patients and, similarly, the EHRs with the GPs. This resulted in the conversation guide prompts sometimes referring to previous information or building on an opinion already voiced.

The patient conversation guide included prompts to the patient about their role in managing their own health; the role of the professionals in their care; perceptions of communication, coordination, and collaboration; and the experience and acceptability of diary keeping. The GP and other professional conversation guides included prompts to the GP or other professional about their role in communication, coordination, and collaboration between professionals; views of barriers and facilitators to effective collaboration; and the experience and acceptability of the methods.

3. Selected portions of the EHR

At completion of the interviews with professionals, EHR database fields containing records of each patient’s care were extracted for the dates of the diary-keeping period. These records included clinical consultation notes, related follow-up records, medications, letters/emails of follow-up from correspondence or referral between general practice (GPs and practice nurses) and other professionals, and all laboratory and radiological results. Personal identifying information was removed from the copies collected.

4. Other forms of written communication

Written communications from the nominated professionals to other professionals not held in the general practice EHR were collected at completion of the interviews. These included email and letter communications. Personal identifying information was removed from all communications.

In accordance with CSOR methodology, the data collected through indirect observation were analysed prior to the collection and analysis of non-observational data (Figure 1). Completed patient diaries were entered into a database. Interviews were transcribed and then analysed by two researchers using inductive, iterative thematic analysis, with the first researcher coding according to topic area, and the second researcher analysing selected topics (e.g. working in partnership, collaboration) for themes that were then discussed and agreed upon by the research team [40]. The selected portions of the EHR and other forms of written communication were entered into databases (the interprofessional interaction data). To ensure rigor of the CSOR methodology of undertaking separate and sequential analyses of multiple methods, all data sources, databases, and analyses remained separate before purposefully integrating the results [41].

The interprofessional research team categorized the interactions according to the 3Cs model proposed by Cohen et al. (developed to explain the nature of integrated care and therefore likely to be applicable to, and explicate, the possible elements of interprofessional, interagency, collaboration) [22]. The categories used were as follows: consultation – being advice-seeking and advice-giving; coordination – being separate, but aligned, care delivery; and collaboration – being shared sense-making and decision-making [22]. These categories were then further tested by corroborating with the interview data.

For the current study, data of two patients who best illustrated the complexity of multimorbidity are presented (Patients 1 and Patient 2). E.M.M. and S.J.M. selected the two patients for analysis, as they met the criteria for seeing other professionals and also agencies in addition to their GP in the 12–13-week diary-collection period. The selection was reviewed and verified by all authors. The remaining two patients, either did not see any other professional/agency or saw only one professional/agency in the 12–13-week diary-collection period. Box 1 shows brief case overviews of Patient 1 and Patient 2, and describes the general practices involved; this summarizes all of the information known about the patients at the project start. The characteristics of the two patients reflect some of the diversity of those with multimorbidity, including age, number and type of conditions, as well as social, economic, and environmental complexity.

Using the health encounter diaries, we identified the professionals/agencies with whom the patients made contact over the data collection period and the number of contacts made with each professional (see Table 2).

Patient 1 used five different methods to contact professionals/agencies who she herself classified as either “health” or “wellness” professionals, the latter usually complementary alternative medicine (CAM) professionals. She recorded 55 contacts with 18 professionals/agencies, with varying numbers of contacts with each.

Patient 2 used two different methods of contact and recorded 17 contacts with six different professionals/agencies with varying numbers of contacts with each.

Schematic maps were created indicating the extent of the contacts with professionals/agencies, as shown in Figure 2.

As well as recording day-to-day individual contacts with professionals/agencies, patients recorded whether or not professionals/agencies mentioned being in contact with anyone else in relation to their care. Patient 1 reported that she was informed that professionals/agencies had been, or planned to be, in touch in 15 of the 55 contacts. Patient 2 reported that she was informed that professionals/agencies had been, or planned to be, in touch in eight of the 17 contacts. There were other occasions where both patients were uncertain if contact was planned or had even occurred.

In the subsequent interviews with the two patients, the interviewer explored how the patients interacted with the professionals/agencies and if they thought that the professionals/agencies communicated with other professionals/agencies. Patient 1 responded to the interviewer’s question, which referred to a previous conversation about “holding all the threads”, and described her role in actively controlling or filtering communication with, and between, professionals involved in her care, depending on the professional’s perceived acceptance of her self-management strategies. Interviewer: “So do you see yourself as, as holding all the threads? … navigate yourself through all this thing?” Patient 1: “Yes. Yes I do. Yes. … I mean I have a superb GP. But I mean I don’t dare mention X [a wellness professional’s name] in her presence, you know. … and I didn’t tell [the GP] this, or the breast surgeon that … I was having vitamin C infusions when, I was having the [oncology] treatment”.

Patient 2 discussed not knowing if her GP (GP2) communicated with the midwife (her nominated professional) and then responded with a somewhat uncertain response (“possibly”) to the interviewer’s query about the possible value of this. Patient 2: “I don’t think they do [talk to each other]”. Interviewer: “Would it be good if they did?” Patient 2: “I guess possibly at stages there would have been times when it could have been useful. But I don’t know what the outcome would have been… except that that my GP and my midwife would know what’s happening”.

An analysis was undertaken of both the contacts and communications recorded in each of the patients’ general practice EHRs, letters, and other correspondence held by the professionals. Based on the total number of professionals involved in the care of each patient (see Tables 2 and 3), there were six recorded contacts between nine of the 24 professionals/agencies for Patient 1, and 22 recorded contacts between 13 of the 14 professionals/agencies for Patient 2.

The most frequent methods of contact between GPs and other professionals/agencies about the patients were via letter (12 times) and electronic methods (11 times via email or secure electronic messaging). There was one internal face-to-face meeting.

In the interviews with the two GPs, the interviewer explored the GPs’ interaction with other professionals or agencies. GP1: “[It’s] by letter [to the specialist], mostly it’s asynchronous, by written correspondence. That’s the most common, with the patient in the middle, sometimes conveying information, it’s quite uncommon for a GP to be actually speaking to a specialist”. Interviewer: “Not a face-to-face?”. GP1: “In fact, some of these specialists, I’ve dealt with them for years, and I wouldn’t have met them”.

In contrast, the nominated professional (physician) working with Patient 1 described attempting to actively communicate with both the patient and the GP, explaining his belief why this was important. Interviewer: “So do you have a particular view on collaboration, then? [Patient 1] thought it was unusual that you included her in the [GP’s] letters, and she appreciated that”. Physician: “… increasingly I think that’s a good way of communicating… it’s pretty straightforward to actually include the patient in the communication, but it hasn’t been the tradition in medicine”.

GP2 expressed a general concern about the type and level of interaction that occurred with Patient 2 and the ways in which to improve interactions. Interviewer: “Has being part of this tiny study… made you think any differently about collaboration?” GP2: “Yes it has actually. Because… I don’t think we document this stuff well, I don’t think it would be clear from the outside that this collaboration was happening. So [it] definitely made me think, if this is your model, then you need to …document when going to work closely with [others]. And we’re going to try and make sure we’ve got those bits of information, being clear that we are collaborating. I’d like to retain some sense of being that coordinating role”. Interviewer: “Mm”. GP2: “[currently] you are just firing out a referral, and getting that expertise to come back… hoping that it will come back to you, so that you can reformulate and move on. … So perhaps being more clear about that, even in the written letter, ‘I want to be kept closely informed about the progress’”.

Although GP2 desired a more collaborative model of interaction, from the nominated professional’s perspective (midwife), in reality there were barriers to this happening. GP2 acknowledged that she had not made contact with the midwife. GP2: “I think I remember talking with [Patient 2] ‘Do you know who you’re going to go to [which midwife]?’. I haven’t had any contact with the midwife… at all, actually”. Interviewer: “And they haven’t sought you out?”. GP2: “No, they haven’t sought me out or fed back to say, ‘It’s all going well’. When I left it to [Patient 2] to contact the midwife, how much information did [Patient 2] tell them about where she’s at with her mental health? So I possibly could have closed that loop by checking in with the midwife”.

Information regarding the nature of interprofessional, interagency multimorbidity care was obtained from EHRs and other forms of written communication from the nominated professionals, and interviews. Analysis of all forms of interprofessional communication data resulted in the formation of spheres of interaction revealing the nature of the interprofessional, interagency care. For Patient 1, there were two connected spheres of interaction (see Figure 3). The first sphere included interaction between some of the professionals/agencies in the care of Patient 1 and GP1. Analysis of the documentation showed a one-way communication of information between each professional/agency and the GP, or the use of ‘report-back’ phrases usually in response to a formal referral letter/email, such as “have filled in the form for [insurance]”. This form of interaction meets the definition proposed by Cohen et al. for consultation (see arrows marked in blue in Figure 3) [22].

The second sphere was between the nominated professional (physician) and other professionals/agencies. Analysis of the interaction showed that, although still mainly unidirectional (in response to contact or referral), it was either responding to and/or inviting further contact. However, in contrast to the consultation form of communication, the written communication used phrases inviting other professionals to align with care, such as, “She will need a repeat of her [name of medicine] which I would be grateful if you could organize”. This form of interaction meets the definition of coordination, as proposed by Cohen et al. (see arrows marked in red in Figure 3) [22].

For Patient 2, two spheres of interaction were identified, but they were only connected by one e-carbon-copied (cc’ed) communication between two of the professionals (see Figure 4). The first sphere involved GP2 and, similar to Patient 1, included multiple, mainly unidirectional, communications between GP2 and a professional/agency, or between a professional/agency and GP2. Analysis of the interaction showed that, except for two instances (see below), these were one-way (in response to contact or referral), indicating consultation (see arrows marked in blue in Figure 4) [22]. However, in contrast to the one-way form of interaction, there were also two interactions in which brief bidirectional exchanges of information occurred. One interaction was a phone call between the practice nurse working with GP2 and the crisis mental health agency (CATT), and was recorded as a file-note in the EHR as, “Phone call to [CATT] to make sure they received the referral and that they are taking care of it”. A letter from CATT to the practice nurse on that same day reported the actions of the CATT in responding to the referral. The second interaction was between GP2 and the maternal mental health (MMH) agency, and was recorded as a file-note in the EHR as, “Phone call to maternal mental health to inform [a clinical event]”. Seven days later, a letter from the MMH was sent to the patient and copied to the GP, acknowledging the phone call and noting the agency had confirmed an appointment time with the patient. These interactions (recorded summaries in the EHR and copies of letters) detail joint decisions and actions to be undertaken by the various parties, and meet the definition of collaboration, as proposed by Cohen et al. (see arrows marked in green in Figure 4) [22].

The second sphere of interaction included bidirectional communications between the nominated professional (midwife) and another agency (MMH), including the involvement of two MMH clinicians (psychiatrist, mental health nurse). Analysis of the written documentation (email) showed a reciprocal exchange of information about the care of Patient 2, with one clinician inviting a follow-up contact, “If you need to get hold of me at any stage you can either call the clinic on [phone number of agency] or my cell [phone number of mental health nurse]”. This interaction meets the definition of coordination, as proposed by Cohen et al. (see arrows marked in red in Figure 4) [22]. However, when this information was augmented by the interview data, it was apparent there had been further unrecorded phone communication between the nominated professional (midwife) and the mental health nurse from the MMH agency in which they had planned a joint meeting together with Patient 2. This interaction meets the definition of collaboration, as proposed by Cohen et al. (see arrows marked in green in Figure 4) [22].

Despite Patient 1 and Patient 2 experiencing complex multimorbidity, where shared decision-making through coordination or collaboration might have been expected, the majority of the individual interactions between the professionals were by nature, consultation alone. There were a few instances in which the interactions with Patient 2 showed separate, but aligned care delivery, indicating coordination, or shared sense-making, decision-making, indicating collaboration (see Figure 4).

Collaboration seemed more likely to occur when professionals phoned each other (with these conversations recorded in the EHR or in an email). The first instance involving GP2 and Patient 2 was in response to a mental health crisis when the GP2 phoned [“tagging”] the counsellor to assess risk. GP2: “[I] needed to get some counselling for her, because she needs to make some decisions quickly around this pregnancy [to] get some support for her mental health. So that was a clear identified need from her side, as well as a fairly obvious one from my side. And then even when she was seeing the counsellor I would be sort of tagging with the counsellor [phoning the counsellor] to make sure, she wasn’t deteriorating …”.

The second instance of collaboration, involving the nominated professional (midwife) and Patient 2, was in response to the midwife foreseeing that she would need to collaborate with the MMH agency in relation to the care of Patient 2. Midwife: “I also made sure that Maternal Mental Health got hold of her, because [there shouldn’t be] a delay in her being seen. … Yes. I chased them up. I did sort of, put a little bit more pressure [on]”. This example of collaboration proved very helpful from the perspective of Patient 2, and when interviewed, she described how she was about to meet with the midwife and the mental health nurse from the MMH agency to form a joint care plan. Patient 2: “There’s a huge overlap with my carers… Like tomorrow I’m meeting with my midwife and my mental health nurse. That immediate plan, post-natally, and to get all that in place, and to have some strategies ready”.

The final stage of analysis combined all data sources (patient health encounter diaries, EHRs, other forms of written communication from the nominated professionals and interviews) to create expanded case syntheses involving Patient 1 and Patient 2 and the professionals.

There are strengths and limitations in this study. The case study design does not purport to produce generalizable results, rather it illustrates problems for the particular cases in their specific health system and community context [31]. The number and variety of data sources used to build each case study in this study is a clear strength [31, 58], and all proved to be necessary when triangulating the various datasets (patient, GP, other professional). Yet, analysing the data sources separately prior to integration and keeping data sources identifiable in the results provides a clear chain of evidence assuring dependability.

Two case studies of those with multimorbidity are provided. While there is no typical profile for such a person, and social circumstance as well as personal attributes and capabilities influence how well people manage [15], ideally, in the future, this selection would be supplemented by others to elucidate a range of experiences. While this study focused on contacts with professionals, future studies should include others in the patient’s social network, including lay carers, as they are known to provide significant support [59].

The challenging observational health encounter diary method is likely to be even more difficult for patients with multimorbidity. It is not surprising that there was some unreliable recording of the daily diary, despite prompts and encouragement from the research team, and this is a limitation. Although GPs may have selected patients they thought would demonstrate ‘ideal interactions’, in-depth analysis showed that these interactions were variable, and the range enabled interactions to be classified according to Cohen et al.’s 3Cs model. The 12–13-week data collection period may not have been long enough to show or make clear any pre-existing collaborative relationships between professionals.