The Web of Science platform is considered one of the most authoritative citation index databases in the world (Wei et al., 2022; Meng et al., 2024). We conducted a systematic search in the SCI-E of the WoSCC database on February 6, 2023. The following retrieval strategy was employed: TS=((Stroke AND “Cognitive dysfunction”) OR “PSCI”). The specific search strategies are recorded in Supplementary material. The literature published between January 1, 2010, and January 31, 2023, was included. After excluding articles written in non-English, we obtained 14,750 original records. After removing the articles not relevant to PSCI and repeated, we finally obtained 1,943 records (Figure 1). Notably, 1,713 research papers were published, while the remaining types of literature included reviews (n = 182), early access (n = 34), meeting abstract (n = 8), and proceedings paper (n = 6).

Extracting the original data from the SCI-E database of the WOSCC inevitably leads to duplicate entries. Therefore, we cleaned the author, country, keywords and other information, and finally imported the cleansed data to VOSviewer (version 1.6.19; Leiden University, Netherlands) and CiteSpace (version 6.2. R3; Drexel University, PA, United States) software applications (van Eck and Waltman, 2010; Chen, 2004) for the subsequent bibliometric analysis. The specific removal process is shown in Supplementary material. Additionally, we utilized Tableau Public software to display the geographical distribution of publications across different countries. Furthermore, we collaborated with the Microsoft Charticulator website¹ to create a string map illustrating international collaboration between clusters. Ultimately, we imported the co-cited references and keyword data analyzed by CiteSpace (version 5.7. R5) into Mapequation² to generate an alluvial flow graph.

Analyzing the annual publication volume provides insights into the evolution of research focus and trends within a specific field. In this study, we finally obtained 1,943 records. Since this study commenced in February 2023, we included publications up to January 2023. As shown in Figure 2A, the annual number of papers shows an upward trend, with the publication volume in 2022 being 4.95 times that of 2010, indicating a growing academic interest in PSCI research. Excluding 2023, the publication trend can be divided into two phases: the first 6 years experienced stable growth, with 599 papers published, while the following 7 years saw a faster increase, with 1,328 papers published. This upward trend suggests a strong and continuing research interest in PSCI, highlighting the field's significant potential for exploration.

We analyzed the number of publications from all countries and highlighted the top five countries during this period (Figure 2B). We found that the annual publication trend in China closely mirrored the global trend, with China consistently leading in publication volume each year. Following China, the USA, the UK, Canada, and Australia also made significant contributions to PSCI research. Among these, the USA exhibited the highest annual publication quantity, although the difference in publication volume between the USA and the other three countries was not substantial. In conclusion, other countries except China may not yet have entered an explosive growth period.

Through VOSviewer visualization analysis, we found that a total of 9,709 authors contributed to this field during the study period. Additionally, we analyzed the authors' collaboration network using CiteSpace software, yielding important metrics such as publication volume and centrality, which are depicted in a bar chart generated by Origin (Figure 3A). The bar chart shows that Vincent C. T. Mok has the most articles (n = 26), followed by Jing Tao (n = 21), Terence J. Quinn (n = 18), and Adrian Wong (n = 17). Among these, Mok, Quinn, and Bordet had the highest centrality (0.05), although these values were relatively low.

Figure 3B presents the author collaboration map based on CiteSpace, which helps identify core authors and the intensity of collaboration between them. The lines between nodes stand for authors' collaboration. The author timezone map (Figure 3C) illustrates the distribution and changes in authors' contributions over different periods. Among the top 15 authors, the earliest contributors were Mok, Wong, and Chu, who laid the groundwork for PSCI research, while Wang, Xiaoling Liao, and Yuesong Pan began publishing in this field after 2021.

Figure 3D displays the co-cited author network. After adjusting the network, we obtained 752 nodes and 2,338 links. As shown in Figure 3D and Table 1, the top five co-cited authors are Pendlebury (n = 525, 0), Nasreddine (n = 300, 0.04), Folstein (n = 249, 0.03), Nys Gms (n = 234, 0), and Hachinski (n = 220, 0.01). Notably, Pendlebury had the highest number of publications but a centrality of 0, suggesting that this author's research content is relatively specialized in PSCI. In contrast, Cordonnier had the highest centrality score at 0.31, despite contributing a comparatively lower number of publications (31).

We adjusted the network to visualize the co-cited authors with a minimum of five citations for time-zone graph analysis (Figure 3E), resulting in 880 nodes and 2,150 links. Pendlebury's node demonstrated the widest yellow ring, suggesting that the literature published by this author received significant attention from scholars in PSCI in 2021. Similarly, Nasreddine's node exhibited the widest orange annual ring, indicating that his publications in PSCI in 2022 were highly cited by many researchers. The differing focus of various scholars over time highlights the evolution of research trends in this field.

Within the PSCI field, 2,356 institutions played various roles during the study period. We used Origin software to create a three-dimensional histogram of institutional information (Figure 4A) and CiteSpace to construct an institutional visualization map (Figure 4B). With a g-index of 15, the network included 267 nodes and 1,026 connections. Combined with Table 2 and pictures, we find that the institution with the highest number of publications was Capital Medical University (n = 69), followed by Institut National De La Sante Et De La Recherche (n = 45), Harvard University (n = 41), University of Oxford (n = 41), and University of Toronto (n = 39). Taken together, although the Chinese University of Hong Kong did not have the highest number of publications (n = 34), it had the highest centrality ranking among all institutions (0.63), indicating a strong influence in this field. The next institutions with high centrality were the University of Glasgow (0.49). Overall, most institutions are closely connected and collaborative.

During the study period, researchers from 76 countries contributed to the PSCI field. Notably, China had the highest publication output (n = 731, 37.6%) and total citations (n = 8,872); however, the country's average number of citations (12.14) and centrality (0.01) were relatively low. The countries with the next highest publication yields were the USA (n = 305, 15.7%), the UK (n = 198, 10.2%), Canada (n = 109, 5.6%), and Australia (n = 99, 5.1%) (Supplementary Table S1; Figures 5A, B). Among these, Canada had the highest average number of citations (34.09), while the UK had the highest centrality (0.28). Canada has garnered significant attention from scholars, and the UK acts as a vital bridge across various PSCI domains. Upon careful observation of Figure 5A, we see that the color width for China in 2022 is notably wide. Combined with Figure 2B, this indicates that China is highly engaged in PSCI research, suggesting that PSCI has become a key focus in Chinese neurology. It is important to note that for countries such as Canada, Italy, and Russia, the red rings indicate periods of heightened publication activity, drawing significant attention from researchers.

Furthermore, we used VOSviewer software to create Figures 5C, D, which represent the network view and overlay view of international collaboration in PSCI research, respectively. Regarding cooperation between countries, we used the Microsoft Charticulator website to create a chord diagram (Figure 5E), allowing us to observe that the thickest cyan arc lies between China and the USA. This relationship indicates that both countries dominate publication output and have the most frequent cooperative exchanges. Notably, the UK had the most connections, suggesting it has the most collaborative partners. Additionally, among the top five countries, China had the least international cooperation, indicating that China should engage in broader international collaborations to explore new areas within PSCI research (Supplementary Table S1; Figures 5C, E).

Using VOSviewer, we constructed a map color-coded to represent the average year of publication by country (Figure 5D). China, Denmark, Portugal, Lreland, Romania, Switzerland, Nigeria, Norway countries had relatively late research in this field. Additionally, we used Tableau Public software to illustrate the geographical distribution of publications across countries (Figure 5F). Among the top 10 countries, the data show a wide geographical representation: China and Japan are situated in Asia; the USA and Canada represent the Americas; Australia symbolizes Oceania; South Korea leads Africa in publication numbers; and Europe is represented by the UK, Netherlands, France, and Germany.

Over the study period, 465 journals published articles related to the PSCI field, citing 7,233 journals. Among the top 10 journals, six were classified as Q1, 3 as Q2, and 1 as Q3 (Figure 6A), indicating the significant influence of PSCI research within the medical field. As shown in Supplementary Tables S2, S3, and Figure 6A, Stroke led in both publication volume (n = 94) and citations (n = 4,391), with a high impact factor (IF), establishing its authoritative influence in PSCI research (Q1, IF 2022 = 8.4). Other leading journals included the Journal of Stroke & Cerebrovascular Diseases (n = 74), Frontiers in Neurology (n = 64), Journal of The Neurological Sciences (n = 37), and PLOS One (n = 34). In comparison, the Archives of Physical Medicine and Rehabilitation had a lower publication output but the highest average number of citations (60.6), indicating significant potential. Conversely, the Journal of Stroke & Cerebrovascular Diseases had high publication and citation volumes but lower average citations per publication and a lower IF, suggesting that the journal has room for improvement in research quality.

Figures 6B–D illustrate the journal collaboration network, average publication time, and density of journal papers in PSCI research, respectively. Using VOSviewer, we identified 46 journals with 10 or more publications, all of which were part of interconnected networks. Among the top five prolific journals, Frontiers in Neurology stood out with a yellow hue, indicating the most recent average publication time and reflecting its status as a preferred journal for many researchers in recent years.

Using the dual-map overlay function of CiteSpace, Figure 6E displays the thematic distribution of journals. On the left are the citing journals in this field, while on the right are the cited journals. Different labels represent the disciplines covered by these journals. The Z-score is related to citation frequency, and the curved paths represent the citation relationships, indicating that publications in the journals on the left may cite publications from the journals on the right. We identified four main pathways in the figure, ranked in descending order by Z-scores (Supplementary Table S4). This analysis reveals that the most frequently covered subjects by citing journals in PSCI are molecular biology, immunology, neurology, sports, and ophthalmology. In contrast, the most cited subjects include molecular biology, genetics, psychology, education, and social sciences.

Figure 6F presents a co-cited journal visualization map, illustrating journals with significant foundational contributions to PSCI research. After adjusting the g-index (k = 15) and trimming the network, we obtained 482 nodes and 1,928 links. As shown in Table 3 and Figure 6F, Stroke is the most co-cited journal (co-citation = 1,690), followed by Neurology (co-citation = 1,111), Lancet Neurology (co-citation = 889), Journal of Neurology Neurosurgery and Psychiatry (co-citation = 780), and PLoS One (co-citation = 640). Notably, there was overlap between the top 10 most productive journals and the top 10 most co-cited journals, including Stroke, Neurology, PLoS One, and Journal of The Neurological Sciences. In terms of co-cited journal centrality (nodes represented by purple circles), Neuroscience Letters had the highest centrality (0.7), followed by Neurochemistry International (0.62), and Neurology (0.47).

When two references are cited together by a third publication, they form a co-citation relationship, which plays a crucial role in the success of the third publication. The higher the citation frequency, the more the reference's research content is considered a foundational knowledge source in the field. The higher the centrality, the more the reference is seen as an inspiration, leading to more derivative research. Among the 53,566 co-cited references, 234 were cited at least 20 times (Figures 7A, B). During the study period, the top 10 most co-cited references were cited at least 105 times, accounting for ~3.36% of all co-cited references, indicating their irreplaceable influence (Table 4). Among them, Pendlebury et al.'s article (Pendlebury and Rothwell, 2009), published in Lancet Neurology (Q1; IF = 48), led the way in terms of citations (co-citation = 359). Other highly influential works included Nasreddine et al. (2005) (co-citation = 300), Folstein et al. (1975) (co-citation = 237), Hachinski et al. (2006) (co-citation =176), and Gorelick et al. (2011) (co-citation = 141) all published in Q1 journals and are considered crucial knowledge sources for future research in the field. Among the top 10 co-cited references, except for Sun et al.'s article, which was not indexed by the SCI, the remainder were published in Q1 journals. Among these journals, two were published in Lancet Neurology, three in Stroke (Q1, IF = 8.4), and the remaining four in the Journal of The American Geriatrics Society (Q1, IF = 6.3), Journal of Psychiatric Research (Q1, IF = 4.8), BMC Medicine (Q1, IF = 9.3), and Journal of Neurology Neurosurgery and Psychiatry (Q1, IF = 11.1).

Using CiteSpace 5.7.R5, we generated a series of co-cited reference networks, which were then input into the alluvial generator. After filtering and adjusting the colors, we obtained an alluvial flow graph of the cited references (Figure 7C). The co-cited references with the longest-lasting streams are listed in Table 5, with the works by Brainin et al. (2015b), Demeyere et al. (2015), and Román and Kalaria (2006) demonstrating the longest flow durations, lasting up to 6 years. The longer the stream lasted, the longer the reference was cited, indicating sustained recognition by PSCI researchers. Additionally, 10 studies had streams lasting 5 years, with Hurford et al.'s (2013) reference having the widest stream, suggesting a significant impact on PSCI research frontiers.

Figure 7D presents the 25 most co-cited references that received significant attention between 2012 and early 2023. By examining the corresponding burst strengths over different periods, we can identify trends in the research focus, helping discover emerging trends in recent years. The figure shows that the studies by Mijajlović et al., Pendlebury et al., Kwon et al., and Zhang et al. have received significant attention in recent years, primarily focusing on risk factors for PSCI (Pendlebury and Rothwell, 2019), objective diagnostic methods (Zhang and Bi, 2020), prognosis (Kwon et al., 2020), and summary reviews (Mijajlović et al., 2017), providing essential foundations for cutting-edge hotspots. Moreover, we found that Pendlebury et al.'s study (Pendlebury and Rothwell, 2009) was of great significance in the field of PSCI (strength = 31.16). Their findings indicated that stroke itself significantly and directly affects the occurrence of dementia, surpassing the influence of vascular risk factors. Furthermore, an increasing number of scholars are exploring new trends through comprehensive reviews.

We used VOSviewer to analyze the keywords extracted from 1,934 publications, which included 5,698 terms. As shown in Figures 8A, B, and Supplementary Table S5, the keyword “dementia” is the most frequently occurring term (588 occurrences), followed by “Alzheimer's disease” (271 occurrences), “brain ischemia” (248 occurrences), “cognition” (246 occurrences), “memory” (229 occurrences), “rehabilitation” (208 occurrences), and “risk factor” (201 occurrences), among others.

During the analysis, 178 keywords appeared at least 12 times and were grouped into four main categories (Figure 8A). Each color represents a cluster, with larger node circles indicating higher keyword frequency. As shown in Figure 8A, we conduct a specific analysis based on the clustering from the largest to the smallest. Cluster one (red) focuses on thecc of PSCI and includes the most keywords, with top terms such as “cognition,” “rehabilitation,” “recovery,” “depression,” “quality of life,” “executive function,” “anxiety,” “aphasia,” “neglect,” “treatment,” and “exercise.” Cluster two (green) mainly involves pathogenesis and covers terms such as “brain ischemia,” “memory,” “model,” “hippocampal,” “inflammation,” “protect,” “oxidative stress,” “injury,” and “plasticity.” The top 10 keywords related to the mechanism are shown in Supplementary Table S6. Cluster three (blue) emphasizes PSCI risk factors, including keywords such as “dementia,” “Alzheimer's disease,” “risk factor,” “vascular dementia,” “TIA (transient ischemic attack),” “brain infarction,” “MRI,” “small vessel disease,” “vascular cognitive impairment,” and “age.” The smallest cluster (yellow) primarily relates to assessment and diagnosis, covering terms such as “MMSE,” “MoCA (Montreal Cognitive Assessment),” “validity,” “diagnosis,” “frequency,” “reliability,” “validation,” “neurological disorders,” “neuropsychology,” and “questionnaire.”

To explore the research topics that have captured attention from 2010 to early 2023, we used CiteSpace in conjunction with the alluvial generator to successfully plot the alluvial flow graph of PSCI's academic keywords (Figure 8C). The keywords with the longest-lasting streams are listed in Supplementary Table S7, with “synaptic plasticity,” “statin,” and “oxidative stress” having the longest streams from 2010 to 2022. This suggests that pathogenesis, treatment, and prevention in PSCI have been critical research topics over the past 13 years. Additionally, the assessment and diagnosis of PSCI have also been widely researched.

To better understand the research trends and hot topics in the PSCI field, we utilized CiteSpace's keyword burst analysis function. The results, as shown in Figure 8D, indicate that the emergence of keywords reflects a rapid increase in frequency over a short period, marking them as hot topics of academic interest. The year 2010 saw the highest citation bursts for keywords such as “vascular dementia,” “cerebrovascular disease,” “vascular cognitive impairment,” “poststroke dementia,” and “validity,” indicating an early focus on PSCI's vascular risk factors and assessment. Research then shifted toward mechanisms, with bursts including “delayed neuronal death,” “global cerebral ischemia,” “rats,” “neurological disorder,” “hippocampus,” “neurons,” and “neurogenesis.” In the past 5 years, keywords such as “synaptic plasticity,” “meta-analysis,” “network,” “tool,” “systematic review,” and “functional connectivity” have frequently appeared.

As shown in Figure 8D, many keywords that appeared around 2010 were related to cerebrovascular diseases. Cerebral infarction caused by CSVD accounts for 25–50% of all strokes, which is relatively higher than that in Western countries (Tsai et al., 2013; Georgakis et al., 2023). CSVD is a crucial pathological basis for cognitive impairment and dementia, significantly increasing the risk of PSD (Lam et al., 2021). We screened the retrieved literature and reanalyzed 56 papers related to CSVD to further explore the focus of CSVD-related studies in PSCI. Using VOSviewer, we analyzed all keywords and identified 23 keywords that appeared at least four times, as shown in Figures 9A–C. As shown in Figures and Table 6, aside from CSVD, “dementia” appears with the highest frequency, followed by “MRI.” MRI is well-known as the most crucial neuroimaging examination method in PSCI (Mijajlović et al., 2017), with typical imaging features including lacunar infarcts, WMHs, CMBs, perivascular spaces, and cerebral atrophy (Du and Xu, 2019; Inoue et al., 2023; Markus and Erik de Leeuw, 2023). Furthermore, imaging marker burdens and severity can reflect cognitive and functional outcomes after stroke (Georgakis et al., 2023). Notably, hypertension is the most important risk factor for non-amyloid CSVD, and primary prevention data have demonstrated that strict control of hypertension may delay cognitive dysfunction (Markus and de Leeuw, 2023). Looking ahead, keyword burst analysis (Figure 9D) suggests that CAA could represent a new research trend in the future.

This study conducted a comprehensive bibliometric analysis of the PSCI field from 2010 to early 2023. The research suggests that the annual number of publications is still on the rise as a whole, indicating that the PSCI field has a broad research prospect in the future. Among the co-cited authors, Pendlebury has been cited the most frequently and is a leading figure in the field of PSCI. China not only has the largest number of publications, but also has the largest number of top 10 institutions, which is mainly related to the increasing importance of PSCI in China in recent years. China has close cooperation with the United States, and the United Kingdom has the largest number of cooperative countries. Nevertheless, there is still a need to strengthen international cooperation in order to achieve wider academic exchanges. Stroke is the journal with the largest number of publications and co-citations, and has a high academic level and international influence.

Two previous bibliometric studies on PSCI visualized relevant data using bibliometric software (Chi et al., 2023; Ou et al., 2023); however, they did not fully discuss a keyword cluster analysis in a comprehensive and systematic manner. In contrast, keyword analysis was at the core of this study. Additionally, we combined the current research hotspots and performed a keyword analysis of CSVD-related articles within the PSCI to thoroughly investigate the significant outcomes between them.

First, to ensure data quality and completeness, the data in this study were obtained exclusively from the WoSCC database. Although the Web of Science database is widely recognized by bibliometric researchers, it cannot cover all relevant studies in this field, which poses a risk of overlooking research from other databases. Additionally, data selection in this study relied on manual filtering to exclude literature unrelated to PSCI, which may introduce bias or omissions. Furthermore, the search language was limited to English publications, inevitably overlooking studies in other languages. Thus, while our search strategy aimed to cover all possible expressions to ensure comprehensive inclusion, there may still be some omissions. Finally, some recently published high-quality articles may have lower citation counts due to their short publication time, necessitating further bibliometric studies in the future to update data and explore new trends and hotspots in PSCI research.

Despite the aforementioned limitations, this study has several advantages. For instance, it combines PSCI with the emerging research hotspot of CSVD for the first time. The visualization results provide valuable information for global PSCI researchers, helping them identify new research directions and hotspots, thereby promoting rapid and efficient development in this field. Additionally, based on our findings and previous clinical studies, we reaffirm the important role of hypertension in PSCI and further explore the association between hypertension and PSCI through its relationship with typical MRI markers of CSVD. These results underscore the importance of early intervention in PSCI and contribute significantly to the diagnosis, treatment, and prevention of the condition. Finally, based on the results of keyword clustering, we systematically and comprehensively discuss the risk factors, pathogenesis, neuropsychological assessment, treatment, and prevention of PSCI. This facilitates a quick understanding of the development process in this field for researchers, provides a scientific basis for clinical practice regarding PSCI, and promotes the overall advancement of the PSCI field.