Community-Based Palliative Care: A Bibliometric and Visual Analysis of Scientific Publications (2015–2025)

Literature Review

The World Organization of Family Doctors recommends delivering PC as a priority through primary healthcare teams in community-based systems (World Organization of Family Doctors, 2014). CBPC supports patients at home or in community facilities through home visits, medical supplies, and telephone follow-up (Hui & Bruera, 2020). Multidisciplinary primary care teams enhance accessibility, feasibility, and cost-efficiency. Compared with institutional care, CBPC is more cost-effective, better aligned with patient preferences, and less burdensome to acute care systems (Hughes et al., 2023; Scott et al., 2024).

Over the past two decades, the evidence base for CBPC has expanded worldwide. Several high-income countries and regions have established relatively mature care models and standardized assessment frameworks (Dale et al., 2009; Hui & Bruera, 2020; Schenker et al., 2021). As an early leader in the modern hospice movement, the UK established a community-led PC model in the 1990s (Coyle et al., 1999), equipping GPs with standardized tools through the Gold Standards Framework (Munday et al., 2007) and integrating CBPC into the national primary care system. The US implemented tiered Medicare payment reforms to create sustainable financial incentives for scalable home-based PC models (Taylor et al., 2018).

China, a developing country, faces compounded challenges: rapid population aging, a growing number of chronic disease patients at the end-of-life, and limited access to PC services (Hu & Feng, 2016). Although policy attention and practical initiatives related to CBPC in China have increased in recent years (H. Lin et al., 2025; Y. Lin et al., 2024), several gaps remain between domestic research and international best practices. These include fragmented research directions, poorly defined core themes, and limited cross-institutional and international collaboration. Further, the lack of systematic reviews has hindered efforts to consolidate the existing knowledge base. These disparities require further validation through comprehensive comparative analyses.

Currently, although a substantial body of research on CBPC exists, bibliometric studies remain scarce. Bibliometrics is a scientific, quantitative method grounded in mathematics and statistics that analyzes specific knowledge carriers (Ninkov et al., 2022). Robust empirical data are essential to clarify the current research landscape, emerging trends, and key thematic areas in this field (Liu et al., 2021). Therefore, herein, we conducted a comprehensive bibliometric and visual analysis of CBPC literature.

We aimed to systematically explore the prevailing themes and evolving trends in CBPC to offer evidence-based insights for informing future research and clinical practice in this field. The analysis aimed to cover publications from 2015 to 2025, using data sourced from the Web of Science Core Collection (WoSCC), to characterize publication patterns and research productivity in this field.

Specifically, we aimed to: (a) provide a comprehensive overview of CBPC research over the past decade, including temporal, geographic, institutional, journal, author, keyword, and co-citation distributions; (b) examine cross-national author collaboration patterns; and (c) elucidate core research themes and their evolutionary trajectories.

Data Source

The WoSCC (Clarivate Analytics, Philadelphia, PA, USA) indexes over 21,000 peer-reviewed journals across disciplines. Known for its citation indexing, it is widely used in bibliometric research. The WoSCC was searched on October 29, 2025. The retrieved records included titles, authors, publication years, journals, countries/regions, keywords, abstracts, and cited references. CiteSpace (version 6.3. R1) and VOSviewer (version 1.6.20) were used to analyze the data and visualize research trends and intellectual structures.

Retrieval Strategy

Data were retrieved from the WoSCC using the following search strategy: (TS=(“Community” OR “Community Hospital” OR “Community Health Centers” OR “Neighborhood Health Centers” OR “Community Health Services” OR “Community Health Care” OR “Community Healthcare” OR “Community Health Nursing”)) AND TS=(“Hospice and Palliative Care Nursing” OR “Palliative Care” OR “Palliative Medicine” OR “Hospice Care” OR “Terminal Care” OR “Hospices” OR “end-of-life care” OR “supportive care”) AND Article (Document Type) AND English (Language).

The search was restricted to peer-reviewed journal articles published between January 1, 2015 and October 29, 2025. Two researchers independently screened the retrieved records by title and abstract. Disagreements were resolved through discussion with a third researcher until consensus was reached.

Data Analysis

Bibliographic characteristics of the selected publications, including year of publication, geographic origin, affiliated institutions, and source journals, were examined using the “Analyze Results” tool in the WoSCC. To illustrate temporal trends and distribution patterns, graphical outputs, including bar charts and line graphs, were generated using Microsoft Excel (version 2016, Microsoft Corporation, Redmond, Washington, United States).

CiteSpace (version 6.3. R1), a Java-based application developed by Chen (2006) for visualizing and analyzing bibliometric networks, was used to identify research fronts and collaboration patterns. Raw data retrieved from the WoSCC were exported in plain text format and saved as download_*.txt. These files were imported into CiteSpace for keyword burst detection and co-citation network analysis. The analysis parameters were set as follows: time span, 2015–2025; years per slice, 1; and selection criteria combining the g-index, k=25; and Top N=50. To evaluate the reliability of the cluster structures derived from co-citation analysis, CiteSpace provides two metrics: modularity (Q) and mean silhouette value (S). A Q value >0.3 suggests meaningful partition of the network into distinct clusters. The S value ranges from -1 to 1, with an average >0.5 indicating coherent and well-matched items within each cluster (Chen et al., 2015). The log-likelihood ratio was applied to interpret the thematic content of each cluster by identifying the associated characteristic and distinguishing terms.

VOSviewer (version 1.6.20), a bibliometric mapping software developed at Leiden University (Van Eck & Waltman, 2010), was used to visualize keyword co-occurrence networks and thematic clusters. Bibliographic data were imported in plain text format. The analysis was configured using the following parameters: fractional counting method and a minimum keyword occurrence threshold of 15. This threshold was applied to exclude low-frequency keywords, ensuring that only representative terms were retained for network construction. For other parameters, default settings were retained. This bibliometric analysis was conducted in accordance with the Preferred Reporting Items for Bibliometric Analysis guidelines.

Ethics Statement

Ethical approval was deemed unnecessary as this study did not involve any human or animal subjects.

Literature Distribution Characteristics

The initial search identified 1,659 publications. After duplicates were removed and titles and abstracts were screened against the inclusion criteria, 1,180 articles were included in the final analysis (Figure 1). They originated from 93 countries/regions.OPEN IN VIEWER

As shown in Figure 2a, annual publication output increased steadily from 2015 to 2025. Figure 2b presents the geographic distribution of research output and the top 20 contributing countries/regions.OPEN IN VIEWER

The institutional distribution of publications is summarized in Table 1. Figure 3 presents the institutional collaboration network in CBPC research, where each node represents a research institution. Intra-regional collaboration is robust, whereas transcontinental ties remain limited.

OPEN IN VIEWER

Table 1.

Top 20 Prolific Institutions (n=1,180)

Ranking	Institutions	Count	Country
1	University of Edinburgh	38	UK
2	King’s College London	36	UK
3	University of Toronto	33	Canada
4	McMaster University	31	Canada
5	Harvard Medical School	27	United States
6	University of Colorado	26	United States
7	University of California, San Francisco	24	United States
8	Monash University	23	Australia
9	Ghent University	23	Belgium
10	Radboud University	23	Netherlands
11	Vrije Universiteit Amsterdam	22	Netherlands
12	The University of Queensland	20	Australia
13	Johns Hopkins University	19	United States
14	Hannover Medical School	19	Germany
15	The University of Sydney	18	Australia
16	University of Cambridge	18	UK
17	University of Leeds	18	UK
18	University of Sheffield	17	UK
19	University of Ottawa	17	Canada
20	Vrije Universiteit Brussel	16	Belgium

OPEN IN VIEWER

Figure 3.

Visualization of the institutional collaboration network

The most productive journal was BMC Palliative Care (Table S1). The most co-cited journal was Palliative Medicine (Impact Factor: 3.9) (Table S2).

Analysis of Authors and Co-Authors

VOSviewer was used to analyze author productivity and collaboration patterns in the field of CBPC (Figure S1). Co-authorship analysis was performed to identify key contributors and their collaborative networks. Four authors were the most productive in the field: Nils Schneider (n=17), Murray SA (n=15), Onwuteaka-Philipsen BD (n=15), and Luc Deliens (n=14).

The author with highest co-citations in this network was Murray SA (175 citations; link strength, 1224), followed by Gomes B (160 citations; link strength, 959). Their strong link strength confirms frequent co-citation, reflecting closely aligned intellectual foundations in CBPC research. Their prominent nodes and robust interconnections establish them as central figures in the co-citation landscape.

Keyword Co-Occurrence and Research Hotspots

A keyword co-occurrence network was constructed using VOSviewer. From an initial pool of 3,452 keywords, 117 terms with >15 occurrences were selected. After a detailed review, no duplicate keywords were identified. Consequently, 117 high-frequency keywords were retained, forming a network with 4,097 links (Table S3). The top five were “palliative care” (n=701), “primary care” (n=251), “cancer” (n=186), “end-stage” (n=165), and “primary health care” (n=152).

Four distinct clusters were identified based on high-frequency keywords, each representing a core research theme (Figure 4). The cluster solution generated by CiteSpace yielded a Q value of 0.415 and a mean S value of 0.742, indicating a statistically acceptable modular structure with good cluster coherence. Cluster 1 (green): CBPC service and practice; Cluster 2 (red): PC and health outcomes among older adults with chronic diseases in the community; Cluster 3 (blue): advance care planning (ACP) and decision-making; Cluster 4 (yellow): barriers and Training of PC in community health workers (CHWs).OPEN IN VIEWER

Burst Detection

Keyword burst detection was performed on 1,180 articles to identify emerging research trends. Figure 5 presents the 30 keywords with the strongest burst strength, arranged by the beginning year for each burst. The timeline is represented by a blue bar, and red segments indicate periods of heightened activity, reflecting duration of each burst. For each keyword, the figure displays the time interval and corresponding burst intensity.OPEN IN VIEWER

Within the analysis period (2015–2025), the earliest keyword to exhibit a burst was “randomized controlled trial,” whereas “discussion” had the longest duration, remaining active for 5 years. In terms of burst strength, the five highest values were observed for “randomized controlled trial,” “directives,” “association,” “older people,” and “interviews,” with strengths of 5.48, 5.27, 5.05, 4.96, and 4.83, respectively. Since 2023, keywords such as “interviews,” “framework,” “family physicians,” “lung,” and “dementia” have emerged as new bursts, suggesting potential future directions in CBPC research.

In addition to the findings presented in Figure 5, CBPC research can be divided into three distinct stages. The initial stage (2015–2018) focused on evaluating the clinical effectiveness and feasibility of CBPC for specific critically ill populations, particularly patients with cancer. During the second stage (2019–2022), research expanded beyond disease-centered interventions to encompass broader aspects, including decision-making in diagnosis and treatment, patients’ psychological and social needs, and support for family systems. The third and most recent stage (2023–2025) emphasizes the systematic integration of PC into the broader healthcare system, with the aim of serving a more diverse patient population while advancing the scientific framework for implementation.

Top Co-Cited Reference Analysis

Analysis of the co-citation network allows examination of development and evolution of specific research domains (Liu et al., 2023). Using CiteSpace, we constructed and analyzed a co-cited network for CBPC. The five most frequently co-cited references are authored by Terry G, Sleeman KE, Kim GM, Sleeman KE, and Clark D. A report predicted that by 2,040 more people in England and Wales will need PC, mainly due to rising deaths from dementia and cancer; this report has the highest centrality (Etkind et al., 2017). References were clustered by title, and the top five clusters identified were “primary care,” “advance directives,” “dementia,” “primary healthcare,” and “quality of care” (Figure S2).

Digital Health Integration: An Emerging Frontier (2023–2025)

Beyond the thematic clusters identified in our analysis, the 2023–2025 period has seen a notable acceleration in digital health integration within CBPC research. Telehealth, remote monitoring, and artificial intelligence (AI) are increasingly shaping the service models and frameworks highlighted herein.

Early empirical studies laid the foundation for this transformation. Udelsman et al. (2019) demonstrated that combining natural language processing with administrative codes enables rapid identification of seriously ill patients who might otherwise be difficult to identify, eliminating the need for labor-intensive manual chart review.

Building on these foundational efforts, recent evidence has broadened the scope of digital health applications. A 2025 systematic review demonstrated that telehealth interventions improve functional status, reduce hospitalizations, alleviate psychological distress, and lessen caregiver burden (Haroen et al., 2025). However, the evidence base remains concentrated in high-income countries, limiting its generalizability. Despite these promising advances, the application of AI in PC remains at an early stage. Although AI has shown considerable potential in areas such as prognostic assessment and medical record support, important challenges remain, including limited validation, a lack of cross-institutional testing, and insufficient transparency (Bozkurt et al., 2025). These issues hinder clinical implementation. Future research should prioritize rigorous external validation, use large patient datasets, and adopt open science approaches to ensure the reliability, safety, and credibility of AI-based tools.

In response to these challenges, digital technologies are reshaping CBPC service models. The EMBED-Care Framework for palliative dementia care digitally combines holistic assessment with clinical decision support, advancing from proof-of-concept to clinical deployment (Davies et al., 2025). Digital advance care planning (DACP) has also progressed, with the Optimal Care recommendations emphasizing interoperability and integration of DACP into routine organizational structures (Allsop et al., 2025).

Collectively, these advances indicate that digital health is reshaping CBPC service delivery. The field has progressed from early predictive models to integrated frameworks that embed clinical decision support into platform-based models. Future research should prioritize evaluating these interventions across diverse settings. A focus on human-centered design is essential to ensure that digital transformation promotes health equity rather than exacerbating disparities.

Main Theme Distribution and Research Frontiers

Analysis of high-frequency keywords helps identify the core research focus areas within the field of CBPC (Gainollina et al., 2025). Burst detection analysis identified several emerging terms, including “dementia,” “lung,” “family physicians,” “framework,” and “interviews,” highlighting potential future directions in CBPC research. Based on analyses of the keyword co-occurrence network and cited references, we identified four core research topics in CBPC research.

Strengths and Limitations

A key strength of this study is the use of established bibliometric tools, including CiteSpace and VOSviewer, to examine publication patterns, collaboration networks, and thematic development in a systematic manner. Despite its rigorous design, this bibliometric study has some inherent limitations. Our search was limited to publications in English indexed in the WoSCC. This ensured consistency in data quality and compatibility with bibliometric analysis tools. This exclusion criterion likely omitted relevant non-English publications and grey literature, thereby compromising the global representativeness of our findings—particularly in regions where English is not the primary language of scholarly communication. Future studies should consider incorporating multilingual databases (e.g., Scopus, PubMed) and grey literature Additionally, the exclusion of non-article publication types, such as reviews, conference proceedings, and books, may have resulted in omission of relevant research contributions. Future studies should incorporate diverse document types to achieve a more comprehensive understanding of the field. Although CiteSpace and VOSviewer are empirically validated, subjective parameter choices and the continuous updating of the WoSCC may introduce variability and database-related bias. Despite these limitations, the literature included herein represents the core body of research in the field and reflects current research hotspots and trends effectively.