Objectives This paper explicates the nature and extent of the networks of care surrounding patients with cancer palliative care needs.
Method Twenty-four patients with 15 different types/sites of cancer were recruited in one city in England, UK. During one in-depth interview patients identified who was ‘involved in their care’ and any known pathways of communication between them. One hundred of these people (35 doctors, 32 nurses, 17 professions allied to medicine, 8 family members and 8 others) were also interviewed. Maps of people/teams and the connections between them for each patient were then reconstructed using social networking software (PAJEK).
Results The 24 patients identified a total of 619 people or teams (mean 26, median 22, range 9–45 per patient) contributing to their care. Selected care network maps are displayed, illustrating the extent and nature of the care networks supporting palliative care patients. Common members of care networks for patients with palliative care needs are revealed, but their individual and unique nature is also apparent.
Conclusions The possible clinical utility and challenges of mapping care networks are discussed. Exploring the care networks surrounding individual patients can be useful for illuminating the extent and complexity of individual patient's care networks; clarifying who is involved and who they communicate with; providing opportunities to see interaction routes that may otherwise be hidden, revealing potentially missing or weak connections; and highlighting overlaps or gaps in provision.
- Supportive care
Statistics from Altmetric.com
The care of people with cancer in England, especially those with palliative care needs, can be complex involving a large team of people.1–3 Communication and continuity of care between them are vital to patient experience, particularly when crossing different geographical locations of care.4–6 National guidance recommends the importance of clear record keeping7 and ‘key workers’ to act as the patient's care co-ordinator,8 although evidence is lacking on who is best placed for this role.9 Individual National Health Service (NHS) trusts have implemented key workers, often a nominated clinical nurse specialist (CNS). They are responsible for continuity of care and support and act as a first point of contact for patients with cancer, liaising with other health and social care professionals (HSCPs) and agencies on behalf of the patient, and ensuring that the patient receives continuity of care when moving between care settings.10 ,11 Knowing who everyone involved in a patients care is and the extent of the communication between them can be challenging.5 ,12 Identifying core palliative care services and ways of improving continuity remain priority questions.13
Explicating the care networks, (defined as the different people and teams contributing to a patient's care, and interactions between them) for patients with palliative care needs could contribute to this understanding. Social network analysis is one way care networks can be analysed.
A scoping review of social networking analysis in healthcare settings identified 52 studies, most descriptive of the networks between health professionals,14 but patients’ networks were specifically excluded from this review. The majority of ‘social network’ research for patients with cancer has focused on on-line support or friendship/family networks in terms of practical or emotional support, or impact on quality of life and survival.15 ,16 A pilot study reporting on written information transmission for six paediatric palliative care patients,17 demonstrates the methodological potential of mapping palliative care networks, but considers only formal, written information sent and received between certain key individuals/departments. Community support networks for carers of people at the end of life used social network analysis, to demonstrate changes in support network size over time for most carers,18 but the position of the patient and HSCPs in the majority of networks was unclear.
Previous research has not been found that: considers the full extent and nature of the care network surrounding a patient with cancer and palliative care needs; includes formal and informal (undocumented, telephone, email or face-to-face) communication; or extends the care network to include both HSCPs and lay, family and patient. Research presented in this paper has not limited the type and extent of the ‘care network’. It incorporates everyone identified by the patient as ‘involved in their care’ in the past 6 months or since diagnosis with cancer (whichever is the shortest time period), and interactions of any type. The aim is to contribute to the literature on palliative care networks by illuminating the extent and nature of example palliative care networks.
NHS Ethical approval and permissions to recruit participants in the selected settings were obtained (REC (Berkshire) reference number: 10/H0505/51). All patients met inclusion criteria (see box 1) and gave informed consent.
Inclusion criteria for patient participants
Patients will be potentially eligible to take part in the study if they are assessed by a health care professional to meet all of the following inclusion criteria:
Diagnosis of cancer
Receiving care in relation to any of the following
Hospital Palliative Care Team
Specialist Palliative Care unit facilities including day, in-patient, out-patient, and community care
Gold Standards Framework (GSF) registers of patients, likely to be in the last year of their life, held by individual general practitioner practices
Resident within the geographical region covered by the hospital trust
Have capacity to provide informed consent
Are at least 18 years of age
Are able, either by themselves or through a representative (eg, a family member, friend or translator) acting as an interpreter or proxy interviewee, to communicate in English with the researcher
Able to cope, physically and emotionally, with the research procedures
Understand that they have cancer, and are able to talk about this with the researcher
Further details on recruitment procedures and qualitative interview data on patient experiences are reported elsewhere.19 Patient recruitment was in one city in England, but through three service routes in order to embrace a range of palliative care needs (see table 1).
Each patient participated in one interview (in hospital, palliative care unit or their home, range 28–169 min, mean 65 min duration). During this interview the researcher asked the patient about all the different people and teams involved in their care since diagnosis with cancer (or in the last 6 months if sooner) and who they all communicated with. Together with the patient, the researcher drew this on a piece of paper, with the patient in the centre of the page surrounded by the patient-identified people and communication routes, thus creating a spider gram or sociogram of their care network. In total, the 24 patients identified 619 people or teams (eg, accident and emergency department, ward nurses, radiotherapy department) engaged in their care. They were recorded as individuals or teams depending on how they were described by the patients. Some of the 619 individuals/teams were in more than one patient network. Accounting for this reduced the overall number eligible to have an interview to 528, of which 272 were identifiable or traceable. With patient consent, invitations to participate in an interview about the care of the patient who named them were sent by letter. If there was no response from HSCPs then follow-up letters were sent, but no further attempts were made to recruit lay/family members who did not respond to the first invitation letter. Of 272 invitations, 100 people (identified by 23 of the 24 patients) were interviewed about the patient's care (range per patient 0–11 people, mean six per patient).
These network member interviews were also audiorecorded either in person or on the telephone (mean 30 min, range 3–87 min).Their contribution to and perspective of the named patient's care network was explored and drawn as a spider gram.
During the mapping processes all participants were free to use any method to recall network members such as diaries, notes etc and a prompt list of potential network members was available in an attempt to get as much detail as possible. Participants were asked who, in addition to them, had the ‘whole picture’ of the network map they had drawn.
Twenty-four patients participated; 15 were woman, and 9 were man. They ranged in age from 48 to 85 (mean 67) years old and had 15 different types/sites of cancer including gastrointestinal tract (n=6), non-small cell lung (n=5), breast (n=5), lymphoma (n=3), genitourinary tract (n=3), and one each of brain, liver, ovary and myeloma. Two patients had two different primary cancers and time since primary cancer diagnosis ranged from 2 weeks to 5 years. At the end of the 2 year project, 13 patients were alive and 11 had died (between 12 days and 8 months after the interview).
Care network members
One hundred members of the 24 patients’ care networks agreed to participate in an interview. Table 2 illustrates them by role and location.
Care network maps were distilled from information gained from the interviews and spider grams and reconstructed within PAJEK,20 a free to download, social networking software package.
A care network map was constructed for each participant individually and then for each patient case. Care networks displayed in this paper are the simplest possible with people/teams being the smallest unit represented as ‘nodes’ and single non-directional lines used to represent any type of connection between them.
Descriptive analysis was through counting the numbers of nodes (people and teams) and lines (communication routes) by KP (with 90–100% agreement on 5 (20%) randomly selected patient maps by NJ). Network visualisation of the sociograms utilises a person's ability to determine important features of the network by looking at the pattern and layout of the sociograms as generated by PAJEK.21 This includes noticing who is incorporated within the network (nodes labelled for role); routes of interaction between them (lines between nodes), key communicators (nodes with many lines); bridges (nodes connecting to two or more other nodes which do not connect directly); groups of nodes (triangles or clusters of connections); network size and density. Density is the number of actual connections divided by the total number of possible connections. The higher the density the more connections there are. If everyone knows and interacts with everyone else in a network the density is 1.0 (100% connected).
Patient participants identified between 9 and 45 people/teams (mean 26, median and mode 22) in their care network. Each care network contained at least one general practitioner (GP) (range 1–5). The next most commonly identified network member was oncologists with up to four oncology doctors identified by 21 patients. Twelve patients had up to five different district nurses and five patients had up to five different home help/care providers coming into their home. A community-based specialist palliative care CNS and a hospital-based tumour type specific CNS were each identified by 10 patients. Eight patients did not identify either type of CNS involvement, while four patients said they had both. Social services and home helps were identified in six and five patient care networks, respectively. The 14 most common people/teams (identified in their network by at least 5 (20%) patients) are listed in table 3.
All 24 patients included family/friends in their care network (range 1–10 people). The majority of patients identified a partner (11 patients) and/or adult child (17 patients), but 5 of the 10 patients who lived alone had no partner or adult child in their care network, although overall the size of care networks appears unrelated to whether the patient lived alone or not. No participants described the full extent of a patient's care network. No patient identified a ‘key-worker’ or someone with responsibility for overseeing the whole network of their care during the interviews, though a ‘key-worker’ was mentioned in some family and HSCP interviews.
Individual patient care networks demonstrate different levels of need and care provision. Patient 14, for example described three tumour specific CNS, one community palliative care CNS, two hospital palliative care team members and five district nurses in their care network; patient 11 included none of these.
Care network maps
Larger networks tend to have more connections. Generally, the patient described networks are not particularly dense (many people in the network are not interacting together), typically utilising less than a third of all possible connections. The trend is for smaller networks to have higher density, indicating more connections between the smaller number of people involved, but this pattern is largely influenced by one patient network who described the smallest network utilising more than half of the possible connections in their network map (density 0.55).
Four example network maps are presented, purposively selected to illustrate uniqueness of patient networks, influences of context and/or potential clinical utility. The first three network maps (figures 1⇓–3) are patient-generated only. The fourth (figure 4) displays the extent and complexity of one patient's palliative care networks by combining patient, family, lay and HSCP maps.
Patient 7 (figure 1) was a 61-year-old man receiving inpatient care in the specialist palliative care unit who usually lived alone. He had a 14-month diagnosis of non-small cell lung cancer and was interviewed 4.5 months before his death. He identified 29 different people and teams involved in his care network, largely related to his current inpatient care. Five of these were GPs from the same surgery. He perceived no interaction between the GPs, while triangles and clusters demonstrate his knowledge of interaction among other network members. His sister is portrayed as a key communicator interacting with family, neighbours, pharmacy, GP surgery and ward clerk and bridge between the GP surgery and the palliative care unit.
Patient 7's map illustrates potential clinical utility, that network size does not indicate network quality or patient need, and context is important. If patient 7 was discharged home, then his care network is small and possibly inadequate, but this might be why he was an inpatient, community care packages could be organised, he may not need or want additional support at home, or he may not go home, for example.
Patient 23 (figure 2) was an 85-year-old woman living at home with her husband, interviewed 1 month after diagnosis of breast cancer and 6.5 months before her death. Her network map was the smallest in the study (nine people) reflecting the context of newly diagnosed and at home, but also had the highest network density, compared to other patients. Higher density (more connections) between network members may be important for perceived support and continuity of care, but a smaller and newer network could have been easier to recall in the interview.
Patient 23 perceived no direct communication between the CNS and GP, indicating this occurs through bridges (husband and patient). Some family (husband and daughter) were key communicators, interacting directly with HSPC and not only among family members as the son and daughter-in-law appeared to.
Patient 1 (figure 3) was a 78-year-old woman with oesophageal cancer widowed and living alone. She was interviewed 7 months following diagnosis and was still alive at the end of the project. Her network map shows her daughters act as important connectors outside of the family, unlike her son who is described as connecting only between a daughter and the patient.
Patient 1 identified 24 people in her care network, 11 of whom were interviewed. Each participant generated a network map for their involvement with patient 1 which was combined in PAJEK (figure 4). This begins to illuminate the full extent and complexity of a patient's individual care network, much of which is unseen by other network members.
Explicating the care networks for 24 patients with a range of palliative care needs for cancer has revealed their uniqueness and complexity. It concurs on some of the challenges for patients and those providing care in knowing who everyone is,5 ,12 the large numbers involved,3 and suggesting difficulty for determining the optimal sized team in end-of-life care.22
Drawing the network map for individual patients brings to attention any potential gaps (eg, no CNS, family, social services); possible duplication of provision (eg, several GPs, CNS or district nurses); unknown or weak connections between network members (eg, bridges provided solely by patient or family members) and key communicators with lots of connections across the network.
This research adds to the literature on mapping care networks in palliative care,16 ,17 expanding on this work and other descriptive network mapping studies by including all types of communication even informal undocumented interaction and extending who is considered to be part of ‘the network’ such as family, friends and other non-professionals.
The care network for someone with palliative care needs for cancer are individual, often complex and include different locations, HSCPs, family and lay persons. For someone to have an overview of it all is challenging. None of the 124 participants in this study produced a complete patient's care network. Policy points to each patient with cancer having a ‘key worker’ who is a health professional with the pivotal role in coordinating and promoting continuity of care,8 but no one person was identified by patients as their ‘key worker’. Suggestions for role of key worker often include the CNS, viewed as vital in contributing to excellence in cancer care,23 ,24 but a third of these patients (8 of 24) did not identify any CNS in their care network.
Care network maps highlight key care providers and connections at a glance. Patients, family and HSCP all hold component parts of this network information, but it is rarely, if ever, distilled into one place in its entirety.
Limitations and future research
There are limitations to this research. Findings relate to 24 patients with cancer in England, UK. A pragmatic decision limited the focus to cancer, but care network mapping is appropriate for any diagnosis or palliative care need. Only HSCP and family identified by the patient as ‘involved in their care’ were invited to participate. The network maps presented are static, cover a fixed maximum of 6 months and rely on participants’ understanding, recollection or system of retaining this information, but this paper privileges this perspective rather than using information verified from health records.
Fishing for statistical associations between network features and variables such as cancer type, time since diagnosis, gender, recruitment route, living alone etc is inappropriate because of small numbers and study design, but could be an avenue for future research, especially if focused on one type of cancer or time point.
The relationship of size and composition of networks to care need or quality was not measured in this study. Networks included everyone identified irrespective of role and involvement, but future research and analysis should be more discerning.
PAJEK is too unwieldy for network mapping in practice and labels lack clarity. Future research could consider other software and patient/clinician preferences for developing individualised care network maps as information organisers. This could be explored further for clinical utility in terms of enhancing record keeping,2 ,7 supporting keyworker role or patient self-management,25 ,26 for example.
This paper illuminates the individual and complex nature of care networks for patients with palliative care needs for cancer and illustrates one potentially useful way to represent and explore these as social networks.
The authors are very grateful to the patients, family members and all the other people involved in patients’ care who kindly gave their time to contribute and share their experiences. Thank you to Dimbleby Cancer Care Research Fund for funding this research project. With many thanks to our advisory group members who comprise patients, carers, clinicians and academics, for their time, support and wisdom during all stages of the study. In particular, the authors would like to thank: Helen Langhorn, Pip McMahon, Tricia Moate, Gaye Orr, Alan Parker and Bernadette Waters. Finally, the authors would like to thank the two anonymous reviewers for their suggestions on how to improve this paper.
Contributors All authors contributed to writing and revision of drafts of the paper, study design, analysis and interpretation of the findings. Contributors who do not meet criteria for authorship are listed in the acknowledgements.
Funding Dimbleby Cancer Care Research Fund.
Competing interests None.
Ethics approval REC Berkshire (ref 10/H0505/51).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.