In England, palliative and end-of-life care (PEoLC) refers to care that enables people of all ages to live well in the last years, months, weeks, days, and hours of life, with “end of life” specifically referring to the final year of life. These terms are often combined to describe the broad services that provide PEoLC [1], which is the term used here. PEoLC systems face intensifying pressures, with aging populations driving demand while services expand multidimensionally to address patient needs [2-6], in-person and digital services [3,7-11], diverse care environments [7,12], its changing societal value [13,14], and a person-centered care approach [15]. PEoLC definitions are evolving through "upstream migration," as services integrate earlier into primary and hospital care rather than being limited solely to end-of-life, expanding the range of illnesses included [16,17]. Hospice care, as a core pillar of PEoLC, reflects this expansion, evolving from single-site cancer relief to multifaceted inpatient and community services within regional networks (Multimedia Appendix 1 summarizes these expansions across dimensions). This expansion and complexity have received recognition from both within and outside the field to approach it on a systemic level [18-22]. Current understanding lags behind these developments, risking reactive services rather than proactive redesign.

Systemic design addresses such complexity with place-based, stakeholder-involved mapping for collective understanding and person-centered interventions [23-25]. Participatory system mapping via causal loop diagrams (CLDs) visualizes this by constructing dynamics for discovery and leverage identification, building joint understanding, and leading to greater trust and sustainability of interventions [26,27]. Distinct from expert group model building (GMB), it engages end users (patients or carers) via directional variable maps for whole-system experience perspectives [28-32]. Participatory system mapping, as a flexible method, is more closely aligned to the format of CLDs, which are directional system maps of variables and their causal relations [33]. By making visible the variables, relationships, and gaps that shape hospice care, participatory CLD mapping can support a more holistic understanding of how services, staff, patients, families, and wider networks interact, thereby identifying points where more coordinated and person-centered care may be strengthened.

Health care participatory design principles and approaches that involve stakeholders are increasing in policy prominence and are seen as enriching the research process, producing better outcomes [34-37]. The National Health Service (NHS) has embedded co-design principles in its “Long Term Plan”; however, they risk tokenism, lack standardization and designers in the process, and are unable to adapt to time or ethical barriers [20,24,25,38,39].

Hospices provide an ideal setting for the exploration of participatory design methods and for researching PEoLC dynamics and operations, compared with alternative locations, by providing access to the core of PEoLC operations and its branching specialist services. This study has selected a hospice operating as a branch of a national hospice care charitable organization in the United Kingdom, which provides robust ground for a holistic, granular, and place-based understanding of hospice care as a complex system. The particular hospice was selected based on several factors, including proximity, willingness for interdisciplinary collaboration, a strong research culture, and support for the researchers in navigating the system and engaging stakeholders.

Our objective is to capture hospice system dynamics and preliminary leverage points via participatory CLD mapping while evaluating method suitability. We have devised 3 research questions in order to achieve this objective: RQ1. What key variables and causal interrelationships do stakeholders identify in a hospice via participatory system mapping workshops? RQ2. What preliminary leverage points emerge from the system map? RQ3. How effective are participatory system mapping workshops in capturing hospice dynamics?

This study formed part of a doctoral research project and received ethical approval from the University of Liverpool, Central University Research Ethics Committee (Ref: 10901; approval document and protocol in Multimedia Appendices 2 and 3). Informed consent was obtained from all participants.

Participatory approaches toward system mapping in health care are assessed as overall positive, with methodological challenges and barriers involved [40,41]. Jun et al [41] note the adaptation of action research methods due to the time constraints of stakeholders in time-limited, one-off workshop settings, as well as the recruitment and retention of busy and geographically distributed stakeholders.

Hospice care is treated here as a focal subsystem of PEoLC because it brings together inpatient, community, bereavement, volunteer, and organizational functions, making it a useful microcosm of broader PEoLC system dynamics. Using participatory research methods in this setting introduces advantages and known barriers similar to those in broader health care settings. The advantages include incorporating many voices collaboratively, as well as reflexive and adaptive approaches that can feed into working groups, continuous improvement, and wider service innovation. The barriers include the high cost of implementation; the time required to conduct research and build productive relationships between researchers and participants; ethical considerations around including patients in sensitive topics; and the challenges of initiating and sustaining participation [35,36,41-44]. This focus allows the study to examine how broader PEoLC dynamics are expressed in a service setting where clinical, community, and organizational congregate.

Adopting the recommendations of Hockley et al [42] and Borgstrom and Barclay [44], the researchers attempted to gain managerial buy-in early on, be reflexive and responsive to participants’ needs, and enable equitable engagement during workshops through effective facilitation. The study adopted a participatory system mapping of CLD through an asynchronous workshop format to allow for flexible, enhanced, and inclusive engagement. CLD was selected because its mapping rules were simple, and once learned, a CLD could be easily understood and built upon by other participants in a nonexpert workshop setting.

Four consecutive 1-hour workshop sessions were developed. A design approach was adopted, and the participatory system mapping was framed, introduced, and delivered as part of a wider design workshop series for system stakeholders (hospice staff and patients). The content of the workshops aimed to first increase participants’ literacy and confidence in the meaning, potential value, and role of technology, design, and systems in hospice care, and then conduct participatory mapping of the hospice system with participants using a CLD. The wider design-led approach to first engage with and understand the relevance and importance of systems design in future hospice care is argued to build contextual relevance and offer value and meaning to participants, thereby increasing meaningful engagement.

“Hospice experience quality” was selected as the central variable for cocreation of the CLD by workshop participants, as it was considered a relevant and important area of priority for all hospice stakeholders, including patients and staff, due to its direct impact on the overall care and well-being of individuals at the end of life.

Accessible and jargon-free language was used in all content, and any potentially new concepts were explained. An information poster was designed for each workshop, outlining its remit and summarizing its content. The language, accessibility, and functionality of all print and digital material were assessed by a patient-facing staff member with no previous familiarity with the topic or content, and improvements were made accordingly. Enhanced engagement and interaction were encouraged by displaying the information poster from the previous workshop in a public space within the hospice, with the ability for participants to add to the discussion via a large notepad, Post-it notes, and pens. Table 1 provides a summary of the 4 workshops, outlining content, aims, and activities.

The workshops were conducted over a period of 3 months in a space accessible to staff, patients, and family members within the hospice, facilitated by an experienced design researcher. At the time of running the workshops, the researcher had spent 8 months in the hospice as part of their designer-in-residence role, conducting observations and interviews, and had developed a good level of familiarity with staff and patients.

The workshop format took shape iteratively throughout the research, exploring the most suitable formats in a high-pressure and sensitive health care setting for participants with very limited time and mental space to engage. This required flexibility in a highly responsive, patient-facing workplace with immediate and fluctuating patient care needs.

The first iteration was informed theoretically by the work of Davis et al [46] and empirically by the format of an existing well-attended 1-hour weekly journal club for hospice staff. Staff engagement helped suggest and later verify a 1-hour duration, a set time slot, and a daily in-person physical format as optimal for attendance and interaction. Four daily 1-hour workshops were delivered consecutively over 1 week, at a consistent time. However, workshop attendance revealed several barriers to participation. Overall, 8 staff members were able to engage over the week, 1 staff member attended all workshops, and no patients participated. Feedback collected from the staff and patients, focused primarily on format and timing, is outlined in Textbox 1.

Overall, 27 participants representing 12 diverse and distinct hospice stakeholder groups attended the workshops. The participating stakeholder groups included patients, doctors, health care assistants and nurses, community nurses, volunteers, management, chaplaincy, well-being team, therapy team, research team, maintenance team, and patient and family support team. This was considered a highly representative sample of hospice care stakeholders. Figures 1 and 2 represent the physical and digital versions of the participatory hospice system map, respectively, codeveloped by hospice stakeholders as workshop participants. The system map contains 84 variables with 175 connections and a network density of 0.03, meaning that the network contains 3% of the possible edges (connections) expected in a completely interconnected network, indicating a relatively sparse network structure compared with a fully connected system.

The participatory and iterative mapping process meant that the hospice system map was validated iteratively and incrementally through each round of stakeholder participation. Participants were asked to both add to the map and check whether the existing variables and their connections on the map were correct, complete, and well explained according to their stakeholder view of the hospice system. In the later stages of map codevelopment, a saturation point was reached where participants describing the effect of a variable could find that connection already represented on the map.

CLDs offer the potential for mixed-method analysis, contributing to a composite understanding of complex systems both quantitatively and qualitatively. The findings presented here are preliminary and represent the outcomes of this process rather than fully articulated results. Further review for missed connections and redundancies, alongside triangulation with other methods or literature, is required for a more rigorous causal understanding of the system. Quantitative and qualitative findings at this stage are presented using network analysis and qualitative interpretation.

Network analysis enables the quantification of the structure of a network and provides insights into the influence and importance of variables on and within the structure, as well as the flow of information within the network [47]. Four commonly used metrics in network analysis, that is, in-degree, out-degree, betweenness centrality, and closeness centrality, were captured.

In-degree measures the number of incoming connections for a variable. Variables with high in-degrees are impacted by multiple other variables. An in-degree of 0 means a variable is not impacted by anything in the system. Out-degree measures the number of outgoing connections for a variable. High out-degree variables can influence more in the system. Variables with a 0 out-degree will not directly impact the system. The codeveloped hospice system map demonstrated in-degree variables ranging from 20 to 0 (Figure 3) and out-degree variables ranging from 8 to 0 (Figure 4).

Betweenness centrality measures a variable’s frequency on the shortest path between 2 other variables. Variables with high betweenness have more control over the flow of information and act as key bridges within the network, referred to as “Mediators” [47]. Closeness centrality measures the distance each variable is from all other variables. In general, variables with high closeness can spread information to the rest of the network most easily and usually have high visibility into what is happening across the network [48]. However, centrality measurements are to be viewed critically. As pointed out by Crielaard et al [49], centrality has been found to vary in results with an adjustment of the positioning of variables in mental models especially as it is a directionless measurement in a directioned map. Table 2 presents the top 3 variables with the highest out-degree, in-degree, betweenness centrality, and closeness centrality out of 84 variables identified. The complete table providing full details is available in Multimedia Appendix 4.

According to the network analysis of CLDs, a key variable is considered one that has a high degree of connectivity, influences multiple other variables, or bridges different parts of the system and, therefore, is identified using the “in-degree,” “out-degree,” “closeness,” and “betweenness” measures. Key causal interrelationships are considered ones that form feedback loops, connect otherwise disparate parts of the system (betweenness), or have a significant influence on the system’s behavior over time.

A leverage point is defined as a variable or relationship in the system where a small change could lead to large shifts in the system’s behavior. According to network analysis, these include the highest out-degree, highest closeness, and highest betweenness [47]. Highest out-degree affects a lot in a system; therefore, changing it should see an effect on many elements quickly. The highest betweenness signifies a bottleneck between clusters, and changing it should see greater connections between variables in the system. The highest closeness affects and reflects the health of the whole system, and changes in it should see a reciprocal effect across the system. Table 3 outlines key variables, causal interrelationships (RQ1), and leverage points (RQ2) in the hospice care system as identified through participatory system mapping with stakeholders.

Key variables and their causal relationships, according to the CLD network analysis metrics, are discussed below.

Participatory system mapping, as part of a design workshop series with hospice stakeholders, proved to be an effective method overall for capturing complex system dynamics in a hospice setting. The research resulted in the iterative development of a hybrid multimodal extended asynchronous method, which improved interaction with participants; accommodated a range of participant preferences in engagement style and digital literacy levels; and allowed for spatiotemporal barriers to be elevated to a degree. The method’s strengths lie in its ability to engage diverse stakeholders, capture in-depth perspectives, and adapt to the unique constraints of the hospice environment. However, challenges related to time constraints, engagement with complex visual data, and the need for researcher guidance in novel tasks suggest areas for refinement or further examination. Future applications should consider a balanced approach between face-to-face and online interactions, with careful attention to spatial and temporal factors unique to each hospice setting. Multimedia Appendix 5 outlines the key strengths and challenges of hybrid asynchronous extended design workshops as a method for participatory system mapping with stakeholders in a hospice setting.

Participatory system mapping offers advantages and disadvantages when compared with alternative systemic mapping methods such as interviews [67], GMB [68], and rich picture mapping [69]. While interviews may provide greater depth into individual perspectives, participatory system mapping supports a collective understanding using participants’ own language, with ideas refined through an ongoing cocreation process rather than filtered solely through researcher interpretation, making it useful as a foundation for continued participatory development, such as co-design. Unlike traditional GMB, the asynchronous approach may increase overall engagement, although it may sacrifice some of the insights generated through real-time group discussion. Rich picture mapping differs from CLD by offering a more open and expressive format, but it may provide less direct structure for systemic analysis.

While the network analysis does not directly demonstrate how to improve patient outcomes, it helps identify where communication, coordination, support, and resource pathways may be strengthened within the hospice system, offering a starting point for more holistic and participatory service investigation and development in PEoLC. Overall, participatory system mapping using CLD provides a balance between accessibility, collective insight, and structured representation of system dynamics, making it well suited to system mapping in sensitive health care settings. Multimedia Appendix 6 summarizes the strengths and weaknesses of the method used in this study compared with alternative approaches.

CLD-based system mapping has inherent capabilities and limitations in terms of the information it can provide on the what, how, and why of a system’s variables and interrelations. In terms of generalizability and thoroughness, this participatory system map provides a snapshot of a single hospice system in an urban context in Northwest England and is solely defined based on the existing understanding and implicit assumptions of its participating stakeholders. Future work could attempt research triangulation by mapping the same system through distinct complementary methods, for example, observations and interviews, thereby enhancing validity and advancing methodological granularity and comparison.

In terms of representativeness and completeness, all identified hospice stakeholder groups were successfully engaged in codevelopment of the system map, apart from the patient family members who could not engage in workshops due to multiple factors. The qualitative interpretation of a CLD is inherently subjective and could be influenced by researcher bias; hence, the combination of quantitative and qualitative CLD analysis in this study.

A CLD system map acts as an analytical canvas, enabling open-ended, multivariable, and granular analysis and interpretation of the whole or parts of a system, depending on the questions being asked about that system. This paper explored key variables, interrelationships, and leverage points of a hospice system as a first step, laying the foundations for the use of CLD in understanding complex hospice system dynamics, and further question-specific discussion of findings could be included.

A greater understanding of the importance of variables and connections could be discovered using Barbrook-Johnson and Penn’s [33] suggestion of weighted connections. However, this would introduce greater complexity and time demands for the workshop activities and facilitation, and it should be approached as a trade-off. The researcher-participant dynamic, through the facilitation of the workshops, may have positively influenced the willingness of participants to share sensitive information, with a nonmedical researcher who lacks the authority to implement changes to the organization directly.

This study developed an innovative hybrid, asynchronous multimodal design workshop to engage 27 diverse hospice stakeholders in participatory system mapping, creating a composite representation of a hospice care system consisting of 84 variables with 175 connections. Key methodological contributions include a reflexive and responsive participatory approach and method that was mostly successful in overcoming challenges common to researching in PEoLC environments, thereby developing a potentially replicable model for ethically engaging diverse stakeholders.

Empirical contributions, derived through mixed-method analysis, reveal structural leverage points discovered in funding, a system bottleneck, and the importance of community organizations and external stakeholder relationships to the system. Further conceptual leverage points identified include reimagining stakeholder roles and expanding the principle of “feeling supported” holistically to the stakeholders of the system. These findings reflect current initiatives found in literature, health care practice, and policy [3,7,53,57,62]. These findings reflect sector-wide challenges and developments reflected from a single site; however, with an absence of family member participants, we lack a critical perspective for a more holistic understanding.

In conclusion, this research provides an informed framework for future participatory studies in PEoLC environments and potentially in broader health care contexts. It demonstrates how accessible participatory methods can produce a holistic understanding of complex systems, generating actionable and grounded insights. Future research could explore the replicability and application of this approach across diverse health care settings aiming to create stakeholder-informed system changes.