A Quarter-Century of Chronic Obstructive Pulmonary Disease in the Intensive Care Unit (2000–2025): A Bibliometric Roadmap of Thematic Evolution and Future Frontiers

Introduction

Chronic Obstructive Pulmonary Disease (COPD) is a leading cause of global morbidity and mortality, representing a major and growing public health challenge.^1–5 Its disease course is frequently interrupted by acute exacerbations (AECOPD), which are key drivers of hospitalization, accelerated lung function decline, and death.^6,7 Within the intensive care unit (ICU), AECOPD is a primary reason for admission, often leading to severe respiratory failure and posing complex management challenges that result in prolonged stays and significant mortality risks.^8–10 The management of these patients is multifaceted, encompassing respiratory support strategies, treatment of comorbidities, and prevention of ICU-acquired complications.^11,12

The past quarter-century has seen significant evolution in the care of critically ill COPD patients. The widespread adoption of non-invasive ventilation (NIV) for hypercapnic respiratory failure, supported by robust evidence, marked a paradigm shift by reducing intubation rates and mortality.^13–16 However, challenges persist, including the identification of patients at risk for NIV failure and the optimization of invasive mechanical ventilation and weaning protocols.^17–20 Beyond ventilation, management is complicated by frequent comorbidities and vulnerabilities to complications like ICU-acquired weakness and secondary infections, all impacting prognosis.^21–27

Recent years have introduced new research dimensions. The convergence of precision medicine and data science has spurred the development of predictive models and prognostic biomarkers.^28,29 Furthermore, the COVID-19 pandemic has had a disproportionately severe impact on COPD patients, intensifying the strain on ICU resources and compelling a re-examination of the role of viral infections in AECOPD.^30–34 The rapid thematic expansion post-2019 coincides with the COVID-19 era, highlighting multifaceted complexities such as prevention-oriented strategies and sex-specific cardiometabolic risks, including the proinflammatory interplay between obesity and Long COVID symptoms in women.³⁵ Concurrently, the research focus has broadened to include long-term functional outcomes and the integration of palliative care.^36–38

This explosive growth in research output, spanning pathophysiology, therapeutics, and prognostication, has created a vast and fragmented knowledge base.³⁹ Clinicians and researchers need a clear, objective synthesis to delineate the field’s trajectory, core foundations, and emerging frontiers. While traditional narrative reviews offer depth, they can be subjective and may not capture macroscopic trends.⁴⁰ Bibliometric analysis provides a quantitative, systematic alternative to map a research field by analyzing publication patterns, collaboration networks, and thematic evolution.⁴¹

A targeted search confirms that while bibliometric studies exist for general COPD research, a dedicated analysis focusing specifically on the ICU context is absent. This study aims to fill this gap. Therefore, we will employ bibliometric methods to systematically map the global research landscape on COPD in the ICU from 2000 onward. Our objectives follow a coherent analytical progression: i) to characterize the field’s growth and key contributing countries/regions; ii) to identify its intellectual core through influential institutions, authors, journals, and their collaborative networks; iii) to map the evolution of core research themes and the intellectual structure via keyword and citation analysis; and iv) to detect emerging research frontiers. This structured analysis aims to provide a data-driven overview to guide future clinical practice and research.

Methods

Data Source and Search Strategy

This study employed a bibliometric analysis approach to investigate the research landscape of COPD within the ICU. The Web of Science Core Collection (WoSCC) was selected as the exclusive data source. This decision was predicated on WoSCC’s reputation for its rigorous indexing standards and the high accuracy of its document type classification, which render it an optimal database for comprehensive bibliometric assessments.

A systematic literature search was executed on October 10, 2025, to identify all relevant publications from January 1, 2000, to October 10, 2025. The search was limited to the Science Citation Index Expanded (SCI-EXPANDED) index. To ensure dataset precision and avoid the retrieval of unrelated ICU literature, the comprehensive search query applied to the “Topic” (TS) field was constructed as follows: TS=(“Pulmonary Disease, Chronic Obstructive” OR “Chronic Obstructive Pulmonary Diseases” OR “COPD” OR “Chronic Obstructive Lung Disease” OR “Chronic Obstructive Pulmonary Disease” OR “COAD” OR “Chronic Obstructive Airway Disease” OR “Airflow Obstruction, Chronic” OR “Airflow Obstructions, Chronic” OR “Chronic Airflow Obstructions” OR “Chronic Airflow Obstruction”) AND TS=(“Intensive Care Unit” OR “ICU” OR “Critical Care” OR “Intensive Care”).

Literature Screening Process

The initial search results were subjected to a meticulous screening process based on a predefined set of eligibility criteria. For inclusion in the final analysis, publications were required to meet the following conditions: (1) the document type was limited to “Articles” or “Reviews” to ensure the inclusion of original research and high-quality syntheses; (2) the language was restricted to English; (3) the publication date fell within the specified range from January 1, 2000, to October 10, 2025. Publications not meeting all three criteria were excluded. Discrepancies between reviewers were resolved through discussion. Following the screening process, the full bibliographic records were exported in Plain Text format with “Full Record and Cited References” for subsequent analysis.

Data Analysis and Visualization

The temporal distribution of annual publication output and the productivity of different countries were analyzed descriptively. A simple linear regression model was fitted to the annual publication count to provide a basic visualization of the overall growth trend; it is important to note that this model is used solely for descriptive illustration of the general trajectory, and the moderate fit (R² = 0.783) indicates that publication growth may follow a more complex, non-linear pattern over time. GraphPad Prism (Version 8.0.2) was used for this descriptive visualization. No inferential claims were made based on these trends.

To delineate the intellectual structure and emerging trends, we utilized two specialized bibliometric software packages: CiteSpace and VOSviewer. Prior to analysis, data cleaning was performed to ensure institutional and terminology disambiguation. A customized thesaurus file was employed to merge synonyms and to consolidate variant institution names.

VOSviewer (Version 1.6.18) was employed to construct and visualize bibliometric networks using the full counting method and Association Strength normalization. For the keyword co-occurrence analysis, a minimum threshold of 15 occurrences per keyword was set to ensure network clarity and focus on established themes. The clustering was performed using the association strength algorithm and the default resolution parameter within VOSviewer. It was used for mapping collaborative networks among authors and institutions, and for revealing core research themes.

CiteSpace (Version 6.2.4.R) was used to perform co-citation analysis and identify research frontiers. The analysis was conducted with the following key parameters: a time-slicing length of 1 year per slice (from 2000 to 2025); node types set to Author, Institution, Keyword, and Reference; and selection criteria utilizing the g-index (k = 25). To optimize the visualization, Pathfinder pruning and Pruning sliced networks were applied. The quality of the resulting clusters was evaluated using Modularity Q and Mean Silhouette metrics. The resulting scientific knowledge maps from both software tools were used to comprehensively visualize the findings.

Results

Overview of the Literature Corpus

Our systematic search of the WoSCC database from January 1, 2000, to October 10, 2025, yielded a total of 2,512 publications pertinent to the bibliometric analysis of COPD in the ICU (Figure 1). This corpus was predominantly composed of original articles (n = 2,250; 89.6%) and review articles (n = 262; 10.4%). The collective body of work involved contributions from 96 countries and regions, 4,004 distinct institutions, and 14,378 individual authors, indicating a broad and diverse international research landscape.

Figure 1 Flowchart of the literature search process.

Temporal Publication Trends

The annual publication output has demonstrated a consistent, albeit non-linear, upward trajectory since the year 2000 (Figure 2A). We have demarcated the publication history into three distinct phases. The initial phase (2000–2010) was characterized by slow growth, with fewer than 50 publications per year, suggesting a nascent stage of research interest in this specific domain. Subsequently, the period from 2011 to 2019 witnessed a phase of accelerated growth, indicating that the field was gaining increasing prominence within the scientific community. The most recent phase, from 2019 onwards, has been marked by a rapid expansion in publication volume, culminating in a peak in 2022. This period of accelerated growth overlaps with the global COVID-19 pandemic.

Figure 2 Global temporal trends and geographic distribution of publications. (A) Annual publication output and growth trend. The light blue bars represent the number of articles published each year. The black dashed line indicates the linear regression analysis of the publication growth, showing a sustained upward trend. (B) Temporal evolution of publication volume by top 10 productive countries. Each colored line represents a specific country (refer to the legend for color-country mapping), illustrating the fluctuations in research output over time. (C) Heat map of publication intensity by country and year. The visual encoding utilizes a color gradient from red (low output) to purple (high output) to represent the annual publication density of the top 10 countries from 2000 to 2025.

Contributions by Countries and Institutions

A total of 96 countries and regions have contributed to the literature on this topic. The United States was the most prolific country, contributing 20.10% of the total publications, a figure that significantly surpassed all other nations (Figure 2B and C). The top five most productive countries were the United States, China, France, Italy, and Spain.

In terms of research impact, publications from the United States garnered the highest number of citations at 20,226 (Table 1), although its citation-per-publication ratio (40.05) ranked fifth among the top ten countries, reflecting a high overall quality of its output. China, the second-most productive country with 392 publications, ranked sixth in total citations (8,505), with a lower citation-per-publication ratio of 21.70. The collaboration network analysis (Figure 3A) revealed strong collaborative ties between the two leading countries, the United States and China. Furthermore, the United States demonstrated significant collaboration with France, the United Kingdom, and Italy, whereas China’s primary collaborative links were with Turkey, Australia, and Germany. The United States’ leadership in publication volume, citation impact, and network centrality firmly establishes its leading position in this research field.

Table 1 Table of Publication Volume by Country

Figure 3 Analysis of international collaborations and journal distribution. In all network maps, node size indicates publication or citation frequency, link thickness represents collaboration or co-citation strength, and concentric ring colors (purple to yellow) signify the temporal evolution. (A) Collaboration networks of countries, visualizing global partnerships over time. (B) Collaboration networks of institutions, visualizing organizational partnerships over time. (C) Density map of journal publications. Color intensity represents the concentration of research within specific journal clusters. (D) Journal co-citation network. Highlighting the foundational journals and their interconnections. (E) Dual-map overlay. Paths illustrate citation flows from citing journals (left) to cited journals (right), representing interdisciplinary knowledge transfer.

At the institutional level, 4,044 organizations have published on the topic. Among the top ten most productive institutions, four are located in France, two in the United States, and one each in Spain, Canada, Turkey, and Egypt (Table 2 and Figure 3B). Assistance Publique - Hôpitaux de Paris (AP-HP) led in productivity with 120 publications, which have accumulated 3,907 citations (32.56 citations/paper). It was followed by the Institut National de la Santé et de la Recherche Médicale (Inserm) (103 papers, 1,788 citations, 17.36 citations/paper), Université Paris Cité (83 papers, 199 citations, 2.40 citations/paper), and CIBER - Centro de Investigacion Biomedica en Red (63 papers, 493 citations, 7.81 citations/paper). Our analysis of collaboration patterns indicates a strong tendency for institutions to collaborate with domestic partners. This observation suggests that fostering greater international collaboration could be crucial for breaking down academic barriers and accelerating progress.

Table 2 Table of Publication Volume by Institution

Analysis of Publishing Journals

The top ten most productive and most co-cited journals are listed in Tables 3 and 4, respectively. The International Journal of Chronic Obstructive Pulmonary Disease was the most prolific outlet, publishing 92 articles (3.66% of the total). It was followed by Intensive Care Medicine (86 articles, 3.42%), Respiratory Care (63 articles, 2.51%), and PLOS ONE (62 articles, 2.47%). Among these highly productive journals, Intensive Care Medicine possessed the highest IF of 22.1. Notably, 90% of the journals in this top-ten list are classified in the Q1 or Q2 quartile (Figure 3C).

Table 3 Table of Journal Publication Volume

Table 4 Table of Journal Co-Citations

A journal’s influence on the scientific community can be assessed by its co-citation frequency. The journal co-citation analysis revealed that the American Journal of Respiratory and Critical Care Medicine was the most frequently co-cited journal (1,445 co-citations), establishing it as a foundational source in the field (Figure 3D and Table 4). This was followed by CHEST (1,412 co-citations) and the New England Journal of Medicine (1,283 co-citations). Among the top ten most co-cited journals, The Lancet had the highest IF (88.5) with 1,100 co-citations. All journals in the top-ten co-citation list were ranked in the Q1 quartile, underscoring the high quality of the intellectual base of this research domain.

Thematic Distribution and Citation Pathways

A dual-map overlay of the journal literature was generated to visualize the thematic distribution and citation flow across disciplines (Figure 3E). In this map, the colored trajectories represent citation links, with citing journals on the left and cited journals on the right. Two primary citation pathways were identified. The most prominent path originates from journals in the “Molecular/Biology/Genetics” domain and flows to journals in the “Medicine/Medical/Clinical” domain. A second significant pathway shows a similar flow from the “Health/Nursing/Medicine” domain to the “Medicine/Medical/Clinical” domain, indicating that research in this field is strongly rooted in basic and health sciences and finds its primary application and impact within clinical medicine.

Leading Authors and Co-Citation Analysis of Authors

Among all authors who have contributed to the literature on COPD in the ICU, the ten most prolific individuals are detailed in Table 5. Collectively, these top ten authors contributed 134 publications, which constitutes 5.33% of the entire corpus of literature in this field. Elie Azoulay and Stefano Nava emerged as the most productive authors, each with 16 publications. They were closely followed by Antoni Torres (15 publications), Laurent Brochard (14 publications), and Zuhal Karakurt (13 publications) (Figure 4A).

Table 5 Table of Author Publications and Co-Citations

Figure 4 Analysis of prominent authors and landmark references. For the network visualizations, node size indicates the frequency of publications or citations, and link weight (thickness) represents the strength of collaboration or co-citation. The concentric rings within nodes signify the temporal distribution of these activities (from purple for earlier years to yellow for recent years). (A) Author collaboration network. Illustrating the primary research teams and their cooperative ties. (B) Author co-citation network. Highlighting the influential scholars whose works are frequently cited together. (C) Reference co-citation network. Identifying core literature that provides the foundational knowledge for the field. (D) Cluster view of co-cited references. Shaded areas represent distinct thematic clusters, labeled by their major research focus. (E) Timeline view of co-cited references. Nodes are arranged horizontally based on their publication year, showcasing the longevity and peak periods of core references.

The scholarly impact of authors was further assessed through co-citation analysis (Figure 4B and Table 5). Fourteen authors were found to have been co-cited over 100 times, signifying their substantial influence and high reputation within the research community. The most prominent nodes in the co-citation network correspond to the most frequently co-cited authors, namely L. Brochard (266 co-citations), S. Nava (225 co-citations), and B.R. Celli (190 co-citations).

Analysis of Co-Cited References

The co-citation network of references, constructed with a one-year time slice from 2000 to 2025, comprised 1,311 nodes and 5,325 links (Figure 4C). An examination of the top ten most impactful publications,^{4,11,13,42–48} ranked by co-citation frequency, reveals the foundational works in this field (Table 6).

Table 6 Table of Co-Cited References

The most highly co-cited article was “Outcomes of noninvasive ventilation for acute exacerbations of chronic obstructive pulmonary disease in the United States, 1998–2008” by Chandra D et al, published in the American Journal of Respiratory and Critical Care Medicine. This study addressed the previously unclear national patterns of noninvasive positive pressure ventilation (NIPPV) use and associated outcomes for hospitalized patients with acute exacerbations of COPD. Utilizing data from the Nationwide Inpatient Sample, the authors found a 462% increase in NIPPV use (from 1.0% to 4.5% of all admissions) and a corresponding 42% decrease in the use of invasive mechanical ventilation (IMV) (from 6.0% to 3.5%) between 1998 and 2008. While in-hospital outcomes generally improved, the study identified a growing cohort of patients requiring a transition from NIPPV to IMV. This subgroup exhibited a worsening trend in in-hospital mortality over time, reaching 29.3% by 2008—a risk 61% higher than for patients initiated directly on IMV and 677% higher than for those treated with NIPPV alone. The study concluded that while the increased adoption of NIPPV was associated with improved overall outcomes, the rising mortality in the NIPPV failure group warranted urgent further investigation.

The second-most co-cited reference was the “Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: GOLD executive summary” by Vestbo J et al, also in the American Journal of Respiratory and Critical Care Medicine. This document from the Global Initiative for Chronic Obstructive Lung Disease (GOLD) presented a major five-year revision of global COPD management strategies, integrating extensive research findings. It reaffirmed that spirometry is mandatory for a clinical diagnosis of COPD. A key update was the emphasis on a multi-faceted patient assessment framework incorporating: (1) symptom burden, (2) degree of airflow limitation, (3) history of exacerbations, and (4) comorbidities. This framework stratifies patients into four groups (A, B, C, D) to guide evidence-based non-pharmacological and pharmacological therapies aimed at reducing symptoms and future exacerbation risk. The summary also prioritized the management of comorbidities and introduced a dedicated chapter on COPD exacerbations, solidifying its role as a foundational guide for national and local COPD management policies.

Thematic Evolution and Research Frontiers

Cluster analysis of the co-cited references revealed the temporal evolution of research hotspots (Figure 4D and E). Early-phase research (circa 2000–2010) concentrated on themes such as acute respiratory failure (Cluster #1), the therapeutic use of helium-oxygen mixtures (Cluster #10), extubation strategies (Cluster #12), and serological markers (Cluster #16).

The focus during the mid-phase of the study period shifted towards topics including non-invasive ventilation (Cluster #3), the pneumonia severity index (Cluster #5), invasive pulmonary aspergillosis (Cluster #7), pulmonary disease as a broader category (Cluster #8), the impact of H1N1 influenza (Cluster #9), respiratory distress syndrome (Cluster #11), exacerbation management (Cluster #13), antibiotics (Cluster #14), healthcare utilization (Cluster #15), and patient behavior (Cluster #18).

Currently, the most recent and emerging research frontiers—representing the field’s active hotspots and future trends—are centered around COVID-19 (Cluster #0), the utilization of large clinical databases like MIMIC-IV (Cluster #2), oxygen therapy (Cluster #4), disease severity stratification (Cluster #6), respiratory syncytial virus (Cluster #17), and rehabilitation (Cluster #19).

Keyword Co-Occurrence and Thematic Clustering

To delineate the primary research themes and developmental trajectory of the field, a co-occurrence analysis of keywords was conducted using VOSviewer. The most frequently occurring keywords were “COPD” (645 occurrences), “mechanical ventilation” (402), “outcome” (294), “COVID-19” (260), and “noninvasive ventilation” (248) (Table 7 and Figure 5A and B). After removing non-substantive terms, a network of 176 keywords, each with a minimum of 19 occurrences, was constructed. This analysis revealed six distinct thematic clusters, each representing a major sub-domain of research. Cluster 1 (Red): Clinical Outcomes, Comorbidities, and Risk Stratification. This largest cluster (55 keywords) focuses on patient outcomes and the impact of various factors. Key terms include outcome, risk, age, burden, complication, cost, death, frailty, palliative care, and quality of life, highlighting a strong focus on the clinical and economic burden of the disease. The presence of keywords like COVID-19, influenza, and infection underscores the significant research interest in the interplay between COPD and infectious diseases. Cluster 2 (Green): Respiratory Support and Intervention Trials. Comprising 46 keywords, this cluster is centered on methods of respiratory support. Central themes include mechanical ventilation, noninvasive ventilation, and oxygen-therapy. Methodological keywords such as controlled trial, meta-analysis, and multicenter indicate a focus on evidence-based evaluation of these interventions. Cluster 3 (Blue): Disease Management, Pathophysiology, and Prediction. This cluster of 37 keywords relates to the clinical management and underlying biology of COPD exacerbations in the ICU. It includes terms like management, biomarker, c-reactive protein, inflammation, and sepsis, pointing to research on diagnostic and prognostic markers. The inclusion of MIMIC-IV suggests a growing trend towards using large-scale clinical databases for predictive modeling. Cluster 4 (Yellow): Intensive Care Management and Procedural Outcomes. This cluster of 21 keywords pertains to specific procedures and management strategies within the critical care environment. Keywords such as extubation, sedation, tracheostomy, and weaning define this theme, focusing on the process of liberating patients from mechanical ventilation and related intensive care practices. Cluster 5 (Purple): ICU-Acquired Infections and Diagnostics. This smaller cluster (13 keywords) is specifically focused on infectious complications within the ICU. Key terms include fungal infection, galactomannan, and nosocomial pneumonia, indicating a research emphasis on the diagnosis and risk factors associated with hospital-acquired infections in this vulnerable patient population. Cluster 6 (Light Blue): Patient-Centric Outcomes. This concise cluster of four keywords (chronic obstructive, survival, delirium, patient) highlights a focus on overarching patient outcomes and critical care complications that significantly affect long-term prognosis.

Table 7 Table of High-Frequency Keywords

Figure 5 Evolution of research hotspots and keyword frontier analysis. (A) Keyword co-occurrence network. Node size reflects the frequency of keyword appearance; link thickness represents the strength of the association between keywords. (B) Keyword density map. Visualized through color intensity, where darker or more saturated regions (purple centers) signify higher frequency concentration and research density. (C) Keyword timeline view. Peaks and troughs illustrate the fluctuating attention given to specific research topics over time. (D) Keyword cluster network. Diamond-shaped nodes represent keywords, grouped into colored shaded regions that define coherent research sub-domains. (E) Top 50 keywords with the strongest citation bursts. The horizontal lines represent the total observation period (2000–2025). The red segments indicate the specific duration of the citation burst, reflecting the rapid emergence and popularity of research frontiers.

To visualize the temporal evolution of these research themes, a burst detection analysis was performed using CiteSpace (Figure 5C and D). This confirmed that noninvasive ventilation, invasive pulmonary aspergillosis, mortality, COVID-19, palliative care, intensive care unit, and chronic obstructive pulmonary disease represent the most recent and strongest research hotspots.

Reference Citation Bursts

We identified the 50 most significant citation bursts among the references using CiteSpace, pinpointing publications that have received a sharp increase in citations, thus signaling a paradigm shift or the emergence of a key research front. The reference with the highest citation burst strength was “Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study” by Zhou Fet al This seminal work provided one of the earliest detailed clinical characterizations of hospitalized COVID-19 patients. The study retrospectively analyzed 191 patients and, through multivariable regression, identified older age (odds ratio 1.10 per year increase), higher Sequential Organ Failure Assessment (SOFA) score (5.65), and an elevated D-dimer level >1 μg/mL on admission (18.42) as critical risk factors for in-hospital mortality. Furthermore, the study revealed a prolonged median duration of viral shedding (20 days) in survivors, providing crucial insights for infection control and therapeutic strategies.

All 50 references with the strongest citation bursts were published between 2000 and 2025, underscoring the contemporary nature of the field’s foundational evidence. Critically, 15 of these publications are currently in an active burst state, indicating their continued and rising influence, and suggesting that research on COPD in the ICU will remain a highly dynamic and impactful area.

Keyword Burst Analysis

An analysis of keyword bursts identified 667 terms that have experienced a surge in frequency over time. The 50 keywords with the most potent bursts were examined (Figure 5E). These terms represent the core of the evolving research frontiers and serve as strong indicators of the field’s future trajectory.

Discussion

This study represents the first panoramic, multidimensional, and quantitative analysis of the global research landscape for COPD in the ICU over the past quarter-century. Our bibliometric investigation reveals a field characterized by exponential growth in scientific output, particularly after 2019; a crystallized core of global research leadership centered in the United States, China, and France; and a profound thematic evolution from fundamental ventilatory strategies to patient-centered, multidimensional outcomes. These findings not only quantify the academic efflorescence of this domain but, more critically, provide data-driven insights into its intellectual architecture, identify key inflection points, and chart a course for its future trajectory.

The tripartite growth curve of publications identified in our analysis mirrors major clinical advancements and global health events. The initial period of slow growth (2000–2010) reflects a nascent stage of research focused on establishing the fundamental applications of invasive and non-invasive ventilation in acute respiratory failure.⁴⁹ The subsequent phase of accelerated growth, post-2011, is directly correlated with the surge of high-level evidence confirming the efficacy of NIV.^50,51 The landmark 2012 study by Chandra et al, identified as the most co-cited reference in our analysis, used large-scale data to demonstrate the widespread adoption of NIV in the United States and its positive impact on overall outcomes, undoubtedly galvanizing subsequent research interest. The explosive growth in publications from 2019 onward is strongly associated with the COVID-19 pandemic, as reflected in our keyword burst analysis. However, this surge is likely multifactorial. Concurrent drivers include the global research imperative triggered by the pandemic, the expansion of open-access publishing models facilitating rapid dissemination, the increasing availability and secondary analysis of large-scale clinical databases, and the continued expansion of journal coverage within the Web of Science. The clustering of “COVID-19” as a key research frontier corresponds to the wave of studies that identified COPD as a critical risk factor for severe disease and mortality from SARS-CoV-2 infection.^46,52

Geographically, the United States dominates in publication volume and total citations, a testament to its vast research infrastructure and leading clinical trial networks, cementing its leadership role.⁵³ However, the rapid ascent of China to the second-most productive nation signals a paradigm shift in the global research power structure, although its lower citation-per-publication ratio suggests a future imperative to enhance research impact.⁵⁴ A particularly noteworthy phenomenon is the clustering of influential French institutions, which account for four of the top ten most productive organizations. This may be attributed to robust, state-funded, and systematic critical care research networks (eg., the REVA network) and the global influence of leading figures in respiratory and critical care medicine such as Laurent Brochard and Elie Azoulay.⁵⁵ Our network analysis, however, also uncovers a pattern of predominantly domestic collaboration. In an era of increasing globalized health challenges like pandemics, breaking down these geographic silos to foster broader international, multicenter research consortia will be essential for tackling complex clinical problems such as difficult weaning and post-intensive care syndrome.^56,57

A critical methodological consideration in interpreting the thematic analysis is the handling of keyword synonymy. We acknowledge that terms such as “NIV,” “non-invasive ventilation,” “NIPPV,” and “BiPAP” refer to overlapping clinical concepts. In our VOSviewer analysis, a customized thesaurus file was employed to consolidate such synonymous terms prior to co-occurrence mapping. This consolidation was essential to prevent the artificial fragmentation of a major research theme into multiple, smaller clusters, thereby providing a more accurate representation of the intellectual structure. The clear dominance of the consolidated “non-invasive ventilation” cluster in our results validates this approach.

Beyond identifying core intellectual pillars, examining the most co-cited references and those with the strongest citation bursts is crucial for delineating the field’s knowledge foundation and evolving frontiers. The five most co-cited works represent seminal contributions that have fundamentally shaped clinical practice and research directions in COPD critical care. The study by Chandra et al (2012) provided pivotal real-world evidence on the rising utilization of NIV and the critical problem of NIV failure, cementing non-invasive ventilation as a first-line strategy while directing sustained research interest towards predicting and managing NIV failure.¹³ Concurrently, the iterative GOLD strategy reports, exemplified by the 2013 executive summary, serve as the cornerstone for standardized diagnosis and management, integrating new evidence into a constantly updated framework that guides global clinical practice and trial design.⁴² The landmark RCT by Plant et al (2000) provided the foundational evidence for the efficacy of NIV in AECOPD, establishing its role in preventing intubation and mortality, and catalyzing its widespread adoption beyond the ICU setting.⁴³ A distinct frontier is marked by the work of Grasselli et al (2020), which, though focusing on a new pathogen, became highly co-cited in this field by definitively establishing COPD as a key risk factor for severe COVID-19, thereby linking the established pathophysiology of COPD to outcomes during a novel pandemic and highlighting this patient group’s unique vulnerability.⁴⁴ Finally, the publication introducing MIMIC-IV by Johnson et al (2023) underscores the transformative shift towards data-driven research. As a foundational resource, it enables the application of advanced analytics and machine learning to develop predictive models for outcomes like mortality and weaning failure, representing the move towards personalized, data-intensive critical care medicine.⁴⁵

Collectively, these highly co-cited references trace the field’s evolution from establishing core interventions (NIV), to standardizing care (GOLD), to confronting new global health threats (COVID-19), and finally to embracing big data analytics. This progression highlights a dynamic research trajectory that is both responsive to emergent challenges and proactive in leveraging new methodologies to address enduring clinical questions in COPD management within the ICU.

The most co-cited references encapsulate pivotal shifts in the field. The study by Chandra et al provided crucial real-world evidence on the rising use of NIV and the associated challenge of NIV failure, directing research towards risk prediction for this high-risk subgroup.¹³ The GOLD strategy report represents the foundational, evolving framework for global COPD management, integrating new evidence to guide practice and research.⁴² Plant et al’s landmark RCT provided the definitive evidence for NIV’s efficacy in AECOPD, cementing its role in management and stimulating widespread adoption.⁴³ In contrast, the work by Grasselli et al, though focused on COVID-19, became highly co-cited by definitively establishing COPD as a major risk factor for severe outcomes during the pandemic, highlighting this population’s vulnerability to novel pathogens.⁴⁴ Finally, the introduction of the MIMIC-IV database by Johnson et almarks a contemporary pivot towards data-intensive research, enabling the application of advanced analytics to develop predictive models and ushering in an era of data-driven critical care.⁴⁵ Collectively, these works trace a clear trajectory: from establishing a core therapeutic intervention (NIV efficacy and real-world outcomes) and standardizing care, to addressing emergent global health threats, and finally to leveraging large-scale data for discovery. This progression underscores a field that is both responsive to acute clinical challenges and proactive in adopting novel methodologies to refine patient management in the ICU.

Our thematic evolution analysis acts as a historical map, vividly illustrating the dynamic shift in research focus for COPD in the ICU. Early research clusters, such as “acute respiratory failure,” “extubation strategies,” and “serological markers,” reflect a primary focus on ensuring patient survival within the ICU.⁵⁸ As fundamental life-support technologies matured, the research agenda evolved towards improving the quality of that survival.

This evolution is most powerfully evidenced by the persistent high frequency of “non-invasive ventilation” (NIV) as a keyword and cluster. The research trajectory has progressed from initial trials proving its ability to avert intubation in select patients to a contemporary focus on precisely identifying high-risk candidates for NIV failure and exploring novel modalities like the combination of NIV with high-flow nasal cannula (HFNC).^59–62 Given that NIV failure is associated with higher mortality, the development of predictive models incorporating clinical scores, imaging, and novel biomarkers represents a critical research frontier aimed at preventing delayed intubation and improving patient outcomes.^63,64

Developing in parallel has been a deepening appreciation of COPD as a systemic disease. The emergence of “Clinical Outcomes, Comorbidities, and Risk Stratification” as the largest keyword cluster highlights a paradigm shift from a single-organ (lung) focus to a holistic, systemic assessment. Comorbidities and complications such as cardiac dysfunction, renal failure, ICU-acquired weakness, and delirium are now well-established as independent predictors of both short- and long-term mortality in critically ill COPD patients.⁶⁵ Consequently, the early risk stratification of these vulnerable patients and the implementation of bundled care strategies (eg., the ABCDEF bundle) are recognized as crucial pathways to improving overall outcomes.⁶⁶ The emergence and bursting of keywords like “frailty” and “palliative care” signify an awakening of humanistic concern and an extension of research endpoints.⁶⁷ A growing consensus now emphasizes the importance of initiating early palliative care consultations for severe COPD patients to facilitate effective symptom management and shared decision-making, moving beyond the mere prolongation of life.⁶⁸

The current research frontiers point clearly in several key directions. First, as previously noted, the impact of COVID-19 has been revolutionary. It has not only exposed the profound vulnerability of COPD patients to novel viral infections but has also spurred deeper investigation into the interplay between viruses (including Respiratory Syncytial Virus and influenza) and bacteria in the pathogenesis of AECOPD.^69–71 Second, the application of big data and artificial intelligence is becoming a powerful engine of progress. The appearance of “MIMIC-IV” as a high-frequency keyword is a direct testament to this trend. Applying machine learning algorithms to electronic health records enables the development of prognostic models with superior accuracy compared to traditional scoring systems, opening the door to personalized, precision critical care.⁷² Third, post-ICU management and rehabilitation has become a new hotspot. As ICU survival rates increase, “rehabilitation” has emerged as a key focus. Addressing how to mitigate the long-term impact of post-intensive care syndrome (PICS) on functional status and quality of life through interventions like early mobilization, nutritional support, and respiratory muscle training is a pressing challenge.^73,74 Fourth, precision diagnosis and treatment of infections. The clustering of keywords like “invasive pulmonary aspergillosis” and “nosocomial pneumonia” reflects a high degree of vigilance regarding infectious complications in the ICU. Against the backdrop of rising antimicrobial resistance, leveraging new technologies like metagenomic next-generation sequencing (mNGS) for rapid pathogen identification, coupled with biomarkers like procalcitonin to guide antibiotic stewardship, is crucial for reducing unnecessary antibiotic exposure and improving infection-related outcomes.^75,76

The co-citation analysis in this study illuminates the intellectual bedrock of the field. Foundational works, such as the study on NIV outcomes by Chandra et al and the GOLD guidelines, form the core theoretical and practical basis of current practice. High-impact journals including the American Journal of Respiratory and Critical Care Medicine, Intensive Care Medicine, and CHEST serve as the central hubs for knowledge dissemination, providing clear guidance for junior researchers on where to publish and access cutting-edge information. The primary strength of our study lies in its large scale, long timeframe, and objectivity, offering intuitive visualizations of complex academic networks and dynamic trends that provide a macroscopic guide to the field’s development.

Nevertheless, this study has several limitations. First, our data was sourced exclusively from the WoSCC, which may have omitted relevant publications indexed in other databases like Scopus or PubMed. This limitation may result in the underrepresentation of research from certain geographic regions and communities that preferentially publish in regional or national journals not indexed in WoSCC, potentially affecting the generalizability of our conclusions regarding global collaboration patterns and thematic emphasis. Second, we included only English-language articles, potentially introducing a language bias that overlooks significant contributions from non-Anglophone countries. Third, bibliometric indicators have inherent weaknesses. Citation counts, while a standard proxy for impact, may underestimate the importance of high-quality recent publications that have not had sufficient time to accrue citations. Furthermore, this analysis cannot assess the intrinsic methodological quality of the included studies. Fourth, as noted, keyword analysis is dependent on the accuracy and consistency of author-provided terms. Although we consolidated key synonyms, variability in terminology remains a potential source of bias in thematic mapping. These limitations, however, do not detract from the core value and the reliability of the main conclusions of this study as the first comprehensive bibliometric assessment of this field.

Conclusions

In summary, this systematic bibliometric analysis of 25 years of literature has comprehensively mapped the knowledge structure of COPD research in the ICU. It demonstrates that the field has evolved from a specialized domain focused on basic life support into a multidisciplinary, cutting-edge frontier that integrates data science, precision medicine, rehabilitation, and palliative care. Based on the latest prominent signals revealed by the bibliometric analysis, future research should focus on four key areas. These directions include: leveraging artificial intelligence and clinical big data to advance personalized precision medicine, and conducting in-depth investigations into ICU-acquired infections and immune dysregulation. Concurrently, it is imperative to improve PICS through rehabilitation interventions and to construct an integrated, continuous care model spanning from the ICU to the community. These priority areas directly respond to the urgent clinical needs highlighted in the data analysis—specifically, the shift from merely improving short-term survival rates to enhancing patients’ long-term function and quality of life.