Unveiling the citation classics: Bibliometric analysis and visualization of the top 100 most cited articles on traumatic brain injury

Abstract

Background:

Traumatic brain injury (TBI) remains a major global health issue with high incidence and mortality, particularly in East Asia. However, most high-impact publications still originate from Western institutions. Significant heterogeneity in study designs, populations, and outcome measures limits the development of standardized clinical guidelines. Bibliometric analysis helps map the scientific landscape, highlight trends and gaps, and support more targeted research.

Methods:

A total of 79,608 publications related to TBI were retrieved from the Dimensions database for publications from 2000 to 2025. From this dataset, the top 100 most-cited articles were selected and further analyzed using VOSviewer to map study design, keywords, institutions, authors, and research collaboration networks.

Results:

Frequent keywords included “Mild Traumatic Brain Injury”, “Intracranial Pressure”, and “Death”. Most studies involved human subjects (n = 82), mainly cross-sectional designs. Mild TBI was predominantly investigated using observational studies, with no specific randomized clinical trial. Only 6% of articles specifically focused on pediatric populations. Prominent institutions included Imperial College London and the University of California, Los Angeles. Leading authors were mainly from the UK and the USA. Despite the high burden of TBI in East Asia, most influential studies came from Western countries.

Conclusion:

This bibliometric analysis identifies critical gaps in TBI research, including the absence of randomized controlled trials evidence for mild TBI, underrepresentation of pediatric research, and a geographic mismatch between research output and disease burden. Addressing these gaps can improve evidence-based practice and advance global TBI care.

INTRODUCTION

Traumatic brain injury (TBI) remains a significant global public health concern, contributing substantially to mortality, long-term disability, and reduced quality of life. In 2021, an estimated 20.8 million TBI cases occurred worldwide, with 56.6% classified as moderate-to-severe. East Asia recorded the highest absolute regional case burden, with 4.3 million cases, of which 69% were moderate or severe. This regional pattern has remained consistent since 1990.[25] Despite these figures, existing estimates may not fully capture the true burden of TBI, as population-based studies project global incidence rates ranging from 790 to over 900/100,000 annually, compared with the age-standardized rate of approximately 599/100,000 reported in registry data, highlighting the likelihood of significant underreporting.[8,12,15]

Beyond the epidemiological magnitude, research on TBI is marked by considerable heterogeneity in study designs, case definitions, outcome measures, and patient populations. This complexity hinders the synthesis of evidence and its translation into standardized clinical practice. As such, bibliometric analysis offers a systematic approach to mapping the scientific landscape, identifying influential studies, revealing knowledge gaps, and characterizing collaboration networks. Thus, this study aimed to conduct a bibliometric analysis of the 100 most cited TBI articles from 2000 to 2025, with the following objectives: (1) identify the most influential studies by citation count, (2) analyze temporal publication trends, (3) map collaboration networks, (4) systematically classify studies by design and population, and (5) identify knowledge gaps warranting future research.

METHODS

Database selection and search strategy

This bibliometric analysis utilized Dimensions database (Digital Science and Research Solutions Inc., London, UK), a comprehensive research intelligence platform. The literature search was conducted on July 1, 2025. The search for these articles was conducted using Boolean operators in the title and abstract fields of (“Traumatic Brain Injury” OR “TBI” OR “Traumatic Encephalopathy” OR “Concussion”). Filters applied included publication years 2000 to 2025, document type limited to articles (excluding books, proceedings, and preprints), and publication status limited to published articles only. No language restrictions were applied at the search stage. This search retrieved 79,608 publications.

The Dimensions database was selected for its comprehensive coverage of scholarly publications and open accessibility of citation data, which enhances study reproducibility. However, citation counts may vary across different databases due to differences in indexing practices, and article rankings might differ if alternative databases such as Web of Science or Scopus were used.

Selection of top 100 most cited articles

From the 79,608 retrieved publications, articles were ranked by total citation count as reported in the Dimensions database as of the search date (July 01, 2025). The 100 articles with the highest citation counts were initially selected for screening. Citations were counted without excluding self-citations, consistent with standard bibliometric methodology that relies on database-reported totals. Although the potential influence of self-citations on ranking cannot be entirely excluded, this approach aligns with established practices in citation analysis.

Inclusion and exclusion criteria

To ensure focus on primary empirical research, articles were evaluated against explicit criteria. Inclusion criteria were (1) primary research articles reporting original data, (2) primary focus on TBI as the main research topic, (3) preclinical studies using animal models, in vitro studies, and biomechanical models, (4) clinical human studies including observational, experimental, and epidemiological designs, (5) published between 2000 and 2025, and (6) available in full text for data extraction.

Exclusion criteria were (1) review articles including narrative reviews, systematic reviews, and meta-analyses; (2) clinical practice guidelines and consensus statements; (3) commentaries, editorials, perspectives, and opinion pieces; (4) studies in which TBI was mentioned only as part of broader neurological disease research; (5) conference abstracts without full-text publication, and (6) retracted articles.

The initial 100 most-cited articles were screened for full text by three independent reviewers. When articles meeting the exclusion criteria were identified, they were removed and replaced with the next most-cited article on the ranked list. This process continued until 100 articles meeting all inclusion criteria were identified. A total of 122 articles were screened to identify the final 100. The complete list of the top 100 articles with citation data is provided in Supplementary Table 1. A flow diagram illustrating the selection process is provided in Figure 1.

Figure 1:

Flow chart of the methodology used in the bibliometric analysis. An overview of the publication selection process to identify the 100 most cited articles discussing about traumatic brain injury, including the reasons for exclusion at each stage.

Study classification

Studies were classified into two main categories: preclinical and human studies. Preclinical studies were further categorized into three subcategories. Cell-based studies included in vitro experiments. Animal model studies were subclassified by injury mechanism, including controlled cortical impact (CCI), fluid percussion injury, weight-drop, blast, laser-induced, compression, or other. Kinematic laboratory studies involved biomechanical models.

Human studies were classified using standard epidemiological taxonomy. Randomized controlled trials (RCTs) were defined as studies with random allocation to intervention or control groups. Cohort studies were defined as longitudinal follow-up from exposure or injury to outcome, including both prospective and retrospective designs. Case–control studies were defined as retrospective studies comparing cases and controls for exposure assessment. Cross-sectional studies were defined as single-time-point data collection, including descriptive and analytical designs. Epidemiological reports were defined as population-based surveillance studies and registry data analyses, such as Global Burden of Disease reports and Global Burden of Disease surveillance summaries. Secondary analyses were defined as post hoc analyses of existing datasets or prognostic model development studies. Diagnostic studies were defined as the development or validation of diagnostic criteria, screening instruments, or clinical prediction rules. Other studies were defined as case reports or designs not captured by the above classifications.

Two investigators independently classified each article, with disagreements resolved by consensus discussion. Articles were assigned to only one category based on their primary study design.

Bibliometric network analysis

Bibliometric network visualization and analysis were conducted using VOSviewer (Centre for Science and Technology Studies, Leiden University, Netherlands). Co-authorship analysis generated network maps to visualize collaboration patterns among authors, institutions, and countries. Keyword co-occurrence analysis examined author-supplied keywords and terms extracted from titles and abstracts.

Statistical analysis

Descriptive statistics were calculated for all bibliometric variables. Continuous variables were reported as means with standard deviations or medians with interquartile ranges as appropriate. Categorical variables were reported as frequencies and percentages. Citation counts were reported as raw totals without per-year normalization, as older publications inherently accumulate more citations. This limitation is acknowledged and discussed. No inferential statistical testing was performed, as this study employed a descriptive bibliometric approach.

RESULTS

Keyword analysis

A total of 79,608 publications discussing TBI were published from 2000 to 2025. The 100 most cited articles were collected using the Dimensions AI database and analyzed using VosViewer software on July 1, 2025. Furthermore, an analysis of the keywords with the highest occurrence counts in such papers was conducted, as described in Table 1 and Figure 2. The most frequent keywords were “Mild Traumatic Brain Injury” (81 occurrences), followed by “Intracranial Pressure” (61) and “Death” (61).

Table 1:

Keywords with more than 30 occurrences. Keywords are listed with their total occurrence counts across the top 100 most-cited traumatic brain injury (TBI) publications and their relevance scores. Occurrence indicates how frequently each term appears, while Relevance reflects the specificity of a keyword in identifying the core topics within the dataset.

Figure 2:

(a) Occurrences and relevance of the keywords with more than 30 occurrences. The black dots represent the number of relevances, and the orange bars represent the occurrences of the 27 most frequently used keywords. (b) Network visualization of the keywords with a minimum of 10 occurrences per term. The various colors represent distinct categories determined by the cooccurrence of keywords. (c) Overlay visualization of the keywords. The gradient colors illustrate the average time of keyword appearance, with darker shades indicating earlier periods and lighter shades signifying more recent ones.

Types of study

Among the 100 most-cited articles on TBI, the majority were human studies (n = 82, 82%), followed by preclinical studies (n = 18, 18%), as shown in Table 2 and Figure 3.

Table 2:

Types of study and the number of articles. The 100 most cited TBI articles are categorized into preclinical and human studies. The NOA for each study type is shown to illustrate the overall distribution of study designs.

Figure 3:

Percentages of the types of studies. Overall, the studies were classified into human studies (82%) and preclinical studies (18%).

Preclinical studies were further classified into cell-based (n = 1, 5.6% of preclinical), kinematic laboratory (n = 2, 11.1%), and animal model studies (n = 15, 83.3%) as illustrated in Figure 4a. Animal studies were further classified by injury induction mechanism. The CCI model was the most frequently employed (n = 8, 53.3% of animal studies), followed by fluid percussion injury (n = 2, 13.3%), weight-drop injury (n = 2, 13.3%), blast injury (n = 1, 6.7%), laser-induced injury (n = 1, 6.7%), and compression injury (n = 1, 6.7%), as illustrated in Figure 4b.

Figure 4:

(a) Percentages of each type of preclinical studies. The breakdown of preclinical studies into animal studies (83.3 %), kinematic laboratory studies (11.1 %), and cell-based studies (5.6 %). (b) Percentages of types of animal studies. This breakdown of animal studies was based on the subjects’ mechanisms of injury, with cortical controlled impact being the most common (53.3%), followed by weight-drop injury (13.3%), fluid percussion injury (13.3%), laser-induced injury (6.7%), blast model injury (6.7%), and compression-induced injury (6.7%). (c) Percentages of types of human studies. The breakdown of human studies shows cross-sectional being the most common (22.0%), followed by randomized controlled trials (20.7%), cohort studies (19.5%), epidemiological reports (13.4%), case–control studies (8.5%), secondary analyses (7.3%), diagnostic studies (7.3%), and other designs (1.2%).

Human studies were classified by study design as follows: cross-sectional studies (n = 18, 22.0%), RCTs (n = 17, 20.7%), cohort studies (n = 16, 19.5%), epidemiological reports (n = 11, 13.4%), case–control studies (n = 7, 8.5%), secondary analyses (n = 6, 7.3%), diagnostic studies (n = 6, 7.3%), and other designs (n = 1, 1.2%), as detailed in Figure 4c. The single study classified as “other” was a case report of chronic traumatic encephalopathy. The temporal distribution of these study types is further illustrated in Figure 5.

Figure 5:

(a) Numbers and percentages of types of preclinical studies compared to human studies by 5 years. Human studies were notably more prevalent than preclinical studies across most time periods, particularly between 2005–2009 (89.2%) and 2010–2014 (80.7%). (b) Numbers and percentages of types of human studies by 5 years. With a diverse distribution across study designs, with cross-sectional being the most frequent study, followed by randomized controlled trials (RCTs), cohort, observational, epidemiological, case–control, and others, particularly in the 2005–2009 and 2010– 2014 periods.

Classification of RCTs

Among the 17 RCTs identified in the top 100 most-cited articles, studies were further classified by their primary intervention focus as follows. Eight RCTs (47.1%) investigated critical care interventions, including hypothermia management, intracranial pressure and oxygen monitoring, and fluid resuscitation strategies. Six RCTs (35.3%) evaluated pharmacologic or medical management approaches, including progesterone, tranexamic acid, and amantadine. Three RCTs (17.6%) focused on surgical interventions, specifically decompressive craniectomy.

TBI severity focus across study designs

Classification of human studies by TBI severity focus revealed distinct patterns across study designs. Among the 17 RCTs, 12 (70.6%) explicitly focused on moderate-to-severe TBI (Glasgow Coma Scale score of 12 or lower), with the remaining 5 RCTs also targeting severe populations through interventions such as hypothermia, ICP monitoring, and decompressive craniectomy. Notably, no RCTs in the top 100 specifically targeted mild TBI populations.

In contrast, among observational studies (n = 41, including cohort, cross-sectional, and case–control designs), the distribution was as follows: 16 (39.0%) investigated mild TBI or concussion, 9 (22.0%) examined moderate-to-severe TBI, 5 (12.2%) included mixed-severity populations, and 11 (26.8%) did not specify severity.

Age-specific study distribution

Analysis of study populations revealed that 6 articles (6%) specifically investigated pediatric TBI. Among these pediatric-focused studies, two were cohort studies examining functional plasticity after early brain injury and the epidemiology of postconcussion syndrome in children. Two RCTs investigated hypothermia therapy and decompressive craniectomy in pediatric populations. One was a diagnostic study (PECARN clinical decision rule), and one was a cross-sectional epidemiological study examining inflicted TBI in young children. The remaining 94 articles (94%) focused on adult populations or mixed-age cohorts without a specific pediatric emphasis.

Publication year

Articles were grouped by publication year, as shown in Figure 6 and Table 3. The article titled “Global, regional, and national burden of TBI and spinal cord injury, 1990 –2016: a systematic analysis for the Global Burden of Disease Study 2016”[12], an epidemiological report by James et al., published in The Lancet Neurology in 2019, received the most citations, with 3,921 as of July 1 ^st, 2025.[12] The peak publication period for articles that subsequently achieved top 100 citation status was between 2005 and 2011, with 11 articles published in 2007 representing the highest single-year contribution. The most recent year in which an article reached the top 100 most cited articles was 2020. No articles published between 2021 and 2025 achieved sufficient citations to enter the top 100 during the study period.

Figure 6:

Number of articles by its year of publication. The trend for publication peaked between 2005 and 2011, followed by a gradual decline, with a marked drop in the number of articles after 2015 and no publications observed from 2021 onward.

Table 3:

Number of articles by year, total citations, and citations per paper. Annual trends in publication volume and citation impact are displayed, including the NOA, total citations received, and average citations per paper for each publication year.

Contribution by institutions

The twelve most contributing institutions had at least 4 publications among the top 100 most cited articles [Table 4]. Imperial College London and the University of California, Los Angeles, contributed the most papers (n = 6 each), with 3,275 and 2,845 total citations, respectively, averaging 545.8 and 474.2 citations per paper. An analysis of their network using VosViewer was conducted and is described in Figure 7.

Table 4:

Institutions with the most contributions to the 100 top-cited articles. Contributing institutions are ranked by the number of publications among the top 100 cited articles. Associated data include the country of origin, total citations received, and average citations per paper to reflect institutional influence and research impact.

Figure 7:

Network visualization of the institutions with the most contributions. The different colors represent the distinct clusters of collaborating institutions, indicating patterns of institutional co-authorship and research collaboration within the field.

Author analysis

A total of 1,385 authors contributed to the 100 most-cited articles on TBI. Authors with four or more publications are presented in Table 5, with an overlay visualization shown in Figure 8. David J. Sharp from Imperial College London, United Kingdom, contributed the most top-cited articles, appearing as an author on 6 of the top 100 articles, with 3,275 citations, an average of 545.8 citations per paper. Sharp is followed by Ewout W. Steyerberg, often affiliated with the Erasmus Medical Center, the Netherlands, with 5 publications in the 100 most-cited articles, 4,025 citations, and an average of 805 citations per paper. Andrew I. R. Maas from the University of Antwerp, Belgium, also has 5 publications in the 100 most-cited articles, with 3,416 citations and an average of 683.2 citations per paper. Moreover, 11 other authors contributed four articles each. Of the fifteen authors with more than 3 articles in the 100 most-cited articles about TBI, seven were from the UK, five from the US, and one from Australia, Belgium, and the Netherlands, respectively.

Table 5:

Most contributing authors of the 100 most cited articles. Authors are ranked by the number of top-cited publications in the field of traumatic brain injury. The data include the NOA, total citations received, average citations per paper, institutional affiliation, and country of origin.

Figure 8:

Overlay visualization of the most contributing authors. The gradient colors illustrate the average publication year of each author’s contributions, with darker colors indicating earlier publications and lighter colors representing more recent contributions.

Top journals

A total of 46 journals contributed to the top 100 most-cited articles. Journals with at least three publications are listed in Table 6, and their network visualization is shown in Figure 9. The New England Journal of Medicine had the highest number of publications (n = 11), with 10,491 citations and an average of 953.7 citations per paper. Brain followed with 7 publications, garnering 5,661 citations and an average of 808.7 citations per paper. Meanwhile, the Journal of Neurotrauma contributed 6 publications, with 3,035 citations and an average of 505.8 citations per paper.

Table 6:

Journals of the 100 most cited articles. A summary of the journals that published the 100 most cited articles. The table includes the NOA per journal, total citations, average citations per paper, Journal Impact Factor (JIF) for 2025 and the past 5 years, H-index, and Journal Citation Reports (JCR) partition.

Figure 9:

Network visualization of journals of the most cited articles. The different colors represent distinct clusters of journals that frequently co-cited the same sources, indicating thematic or disciplinary similarities among the journals.

Nature had the highest H-index of 1,442, while both Journal of Head Trauma Rehabilitation and The Morbidity and Mortality Weekly Report (MMWR) Surveillance Summaries had the lowest H-index of 119. The Lancet had the highest Journal Impact Factor (JIF ) and JIF 5 years later, with 98.4 and 106.9, respectively, while Brain Injury had the lowest, with 1.5 and 2.4, respectively. All of the top 20 most contributing journals were in the Journal Citation Reports first quartile (Q1) partition, except for brain injury. The 20 most contributing journals, with their H-index, 2025 JIF, and 5-year JIF, are shown in Figure 10.

Figure 10:

(a) H-index distribution among the top 20 journals contributing to the 100 most cited TBI articles. Nature ranked first with an H-index of 1,442, followed by the New England Journal of Medicine (1,232), The Lancet (936), and Proceedings of the National Academy of Sciences (896). (b) The Impact Factor/IF 2025 and the last 5 years. Number of publications and journal impact factors (IF) among the top 100 most cited TBI articles. Blue bars indicate publication counts, black and red lines represent 2025 IF and 5-year IF, respectively.

Countries with at least two publications

The top 100 most-cited papers were published across 18 countries and are listed in Table 7. Among them, the United States had the highest number of papers at 70, with 50,698 citations, averaging 724.3 citations per paper. The United Kingdom ranked second with 17 papers and 10,965 citations, averaging 645 citations per paper. The Netherlands ranked third with 7 papers and 4,378 citations, averaging 625.4 citations per paper. The map of country co-authorship, showing the number of articles and density, is presented in Figure 11.

Table 7:

List of countries with at least two articles. A breakdown of countries that contributed at least two articles among the 100 most cited, showing the NOA, total citations, and average citations per paper.

Figure 11:

(a) Map of countries by number of articles contributed. The darker colors in picture (a) represent the countries with the higher number of publications. (b) The density visualization of the contributing countries. The brighter colors in picture (b) indicate higher publication density, with the United States showing the most substantial contribution.

DISCUSSION

This bibliometric analysis of the 100 most-cited papers on TBI from 2000 to 2025 reveals patterns in research methodology, publication trends, and knowledge gaps that have shaped the field over the past quarter century.

One of the most prominent observations is the disparity in study designs across TBI severity categories. All 17 RCTs targeted moderate-to-severe TBI populations, with no RCTs specifically designed for mild TBI. In contrast, observational studies predominantly investigated mild TBI (39.0% of cohort, cross-sectional, and case–control studies). This pattern likely reflects both ethical and practical challenges inherent in enrolling and randomizing patients with mild TBI, who exhibit subtle and transient symptoms.

Across these RCTs, endpoints and reporting frameworks were various. Many of these differences result from heterogeneity in inclusion criteria, populations included, and the use of dichotomized outcomes such as the Glasgow Outcome Scale. These differences complicate direct comparisons across studies and highlight the importance of adopting standardized reporting frameworks for outcomes, thereby enhancing external validity and enabling robust meta-analyses. Across 17 RCTs, eight focused on critical care,[2,4-6,11,16-18] six on medical management, and three on surgical interventions, including decompressive craniectomy.[7,10,20] Among the six that focused on medical management, four of them utilized the administration of progesterone,[19,22-24] one of them utilized tranexamic acid,[21] and the other utilized amantadine.[9]

Another interesting finding is that of the 100 most cited articles, only 6 specifically investigated or observed TBI in the pediatric population.[1,3,11,13,14,20] This relatively small proportion is particularly striking given the distinct pathophysiology of the developing brain, age-specific injury mechanisms such as inflicted trauma in young children and sports-related injuries in adolescents, and different outcome trajectories compared with adults. The underrepresentation may reflect several factors, including smaller patient populations that limit statistical power, ethical challenges in pediatric research, and potentially lower submission rates to the highest-impact journals.

Geographically, the majority of the most-cited studies were not produced at institutions in the Western Pacific and East Asia, where the highest incidence of TBI was reported, but in the United States and Europe. This research-burden mismatch likely reflects multiple factors, including differences in research infrastructure and funding availability, language barriers that affect international dissemination, and the continued prominence of Western-based high-impact journals. This geographic disparity has policy implications for high-burden regions. Countries in Asia, including Indonesia, bear substantial TBI-related healthcare costs yet remain underrepresented in high-impact research. Strengthening regional research capacity and fostering international collaborations could help generate context-specific evidence addressing local challenges, such as predominant injury mechanisms and healthcare resource availability.

An intriguing phenomenon is the absence of top-cited novel publications between 2021 and 2025. While citation accumulation inherently favors older publications, this recent absence is particularly pronounced. This is primarily attributable to citation lag, a well-recognized bibliometric phenomenon in which recently published articles require time, typically 5–10 years, to accumulate citations comparable to those of older publications. This temporal pattern coincides with the COVID-19 pandemic period, during which substantial shifts in research priorities and publication patterns occurred across biomedical fields. However, we acknowledge that this observation should be interpreted with caution, as formal analysis of the pandemic’s impact was beyond the scope of our study. Future bibliometric studies will be needed to determine whether recent articles eventually achieve citation classic status or whether this represents a more fundamental shift in TBI research productivity.

Among preclinical studies, a clear preference emerged for the CCI model, which was used in 8 of 15 animal studies (53.3%). This preference reflects the model’s reproducibility, standardization across laboratories, and established utility for mimicking focal contusional injury patterns observed in human TBI. Human studies consistently achieved higher citation counts than preclinical studies, likely reflecting both the clinical applicability of human research and the persistent translational challenges in TBI research.

This study has several limitations. First, using citation count as our primary selection criterion inherently favors older publications, which have had more time to accumulate citations. Still, we did not apply per-year normalization. Second, we focused exclusively on primary research articles and deliberately excluded reviews and meta-analyses. Third, our analysis was limited to publications indexed in the Dimensions database, and citation counts may vary across databases due to differences in indexing practices. Fourth, the absence of recent citation classics may partially reflect pandemic-related factors, but this remains speculative without formal analysis. Fifth, self-citations were not excluded from citation counts, which may have influenced article rankings to some extent.

CONCLUSION

This bibliometric analysis of the 100 most-cited articles on TBI from 2000 to 2025 highlights key trends and gaps in the field. Observational studies dominate research on mild TBI, whereas RCTs are more common in moderate to severe cases, often with heterogeneous outcomes that hinder comparability. Despite its distinct characteristics, pediatric TBI remains underrepresented among citation classics. Although East Asia bears the highest TBI burden, most highly cited studies originate from Western countries, reflecting global research disparities. Moving forward, priorities for TBI research should include well-designed randomized trials for mild TBI, universal adoption of standardized outcome reporting, increased investment in pediatric TBI research, and strengthened research capacity in high-burden regions.

Ethical approval:

The Institutional Review Board approval is not required.

Declaration of patient consent:

Patient consent not required as there are no patients in this study.

Financial support and sponsorship:

Nil.

Conflicts of interest:

There are no conflicts of interest.

Use of artificial intelligence (AI)-assisted technology for manuscript preparation:

The authors confirm that there was no use of artificial intelligence (AI)-assisted technology for assisting in the writing or editing of the manuscript and no images were manipulated using AI.

Supplementary data available on:

https://doi.org/10.25259/SNI_837_2025

Disclaimer

The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Journal or its management. The information contained in this article should not be considered to be medical advice; patients should consult their own physicians for advice as to their specific medical needs.