Global status and trends in heart failure with preserved ejection fraction over the period 2009-2020

Abstract Background: Heart failure with preserved ejection fraction (HFpEF) comprises about 50% of the cases of heart failure (HF), but so far there is no effective treatment strategy. This study used bibliometric methods to analyze the scientific literature on HFpEF in 2009 to 2020, and evaluate the global scientific output of HFpEF research, in order to explore the research status and trends in this field. Methods: Documents about the HFpEF research published in 2009 to 2020 were retrieved from Science Citation Index Expanded (SCIE) in Web of Science. This study used bibliometrix R-package, VOSviewer, and CiteSpace to conduct the bibliometric analysis. Results: A total of 1971 documents (1508 articles and 283 reviews) were retrieved to construct the local HFpEF literature collection for analysis. The number of annual documents had increased year by year in general, from 24 to 353. Relevant documents were mainly written in English, and mostly focused on the field of “Cardiovascular System Cardiology.” USA ranked first in the relevant countries/regions with most documents, and the leading affiliation was Mayo Clin. Shah SJ was the most productive author, while Borlaug BA ranked highest among the local cited authors and G-index. Circulation was the most local cited source, while Eur J Heart Fail published the most documents and was rated as the top source in terms of G-index. “Paulus WJ, 2013, J Am Coll Cardiol” was the top local cited document within the local HFpEF literature collection, while “Owan TE, 2006, New Engl J Med” outside the local HFpEF literature collection was the most local cited reference. The keywords such as “mortality,” “dysfunction,” “diagnosis,” “outcomes,” and “diastolic dysfunction” were most frequent, while “hemodynamics,” “comorbidity,” “myocardial infarction,” “inflammation,” and “phenotype” indicated research frontiers or emerging trends. Furthermore, this study also found some deeper bibliometric relationships through bibliographic networks. Conclusions: Due to the multi-dimensional bibliometric analysis, this study shows a wide view of scientific productivity related to HFpEF, and provides valuable guidance for researchers interested in HFpEF, assisting them in understanding the research status, identifying potential collaborators, discovering research hotspots and frontiers, and conducting more in-depth research.


Introduction
Heart failure (HF) is the terminal stage of various cardiac diseases, with significant morbidity and high mortality worldwide. [2]Heart failure with preserved ejection fraction (HFpEF) is the most common type of HF, and it accounts for more than half of the cases of HF. [1,[4][5][6][7][8][9][10][11][12][13][14] HFpEF is characterized by normal or near normal left ventricular ejection fraction, [15] so the prognosis of HFpEF patients is often falsely considered to be better than that of heart failure with reduced ejection fraction (HFrEF) patients. [3][15][16][17] In fact, the etiology, pathophysiology, and management of HFpEF and HFrEF are very different. [1,18]In the past 2 decades, the standard treatment of HF has been popularized, which has significantly reduced the hospitalization rate and improved the survival rate of patients with HFrEF. [1]HFpEF currently lacks a successful treatment strategy to improve prognosis.This is attributed to incomplete understanding of HFpEF pathophysiology, patient heterogeneity, and mismatch between treatment mechanisms and major pathophysiologic processes. [19]HFpEF is associated with a variety of medical complications and pathophysiological pathways or clinical phenotypes, accompanied by multiple organ system disorders. [1,12,24]The important clinical phenotypes of HFpEF include diabetes, hypertension, lung disease, chronic kidney disease, obesity, and so on. [12]ersonalized techniques based on individuals' phenotypic traits, diagnosis, and pathophysiology stratification are needed to replace "one-size-fits-all" therapy for HFpEF. [19,22,25]HFpEF, in particular, should be handled differently depending on its phenotypes. [3,19]Only addressing phenotypic diversity of HFpEF can tailor more personalized therapies for specific HFpEF phenotypes. [20]In summary, HFpEF represents one of the largest unmet clinical needs in the current cardiovascular medicine, [2,25- 27] so there is an urgent need to develop new treatments for patients with HFpEF. [1]linical researchers should extensively collect literature and accumulate relevant professional knowledge in the direction of related disciplines to ensure the innovation, scientificity, and non-repetitiveness of the clinical research topics.Bibliometrics is a widely accepted research method, which has been carried out in clinical research topics.Based on statistics and visualization technology, bibliometrics makes quantitative and qualitative analysis on the published academic literature in a specific clinical research topic, describes knowledge structure and tracks development trend.Bibliometric analysis usually includes the following steps: putting forward a research question, retrieving databases, collecting and analyzing data, drawing maps, preparing, and publishing the report. [28]Some bibliometric tools have been widely used by researchers to evaluate the impacts of documents, contributions of authors/ affiliations/countries, patterns of authorship, and the main directions of future research in a clinical research topic.Clinical researchers should develop search strategies, retrieve relevant documents from literature databases, and use bibliometric methods to conduct multifaceted bibliometric analysis of these documents to find clinical research hotspots and determine research directions in a purposeful and targeted manner.At present, there are some bibliometric studies in the field of cardiovascular diseases, [29][30][31][32]45] one of which is on HFpEF. A biliometric study conducted by Edlinger et al [45] showed a great increase in research on HFpEF in the United States, Japan, and continental Europe, and identified RAAS-blockade and 5-phosphodiesterase-inhibition as a therapeutic trend in HFpEF.This study will take the bibliometric analysis of HFpEF a step further to provide a more comprehensive perspective for clinical researchers.

Methods
This study retrieved literature collection from Web of Science.Web of Science (WoS) is one of the world's most trusted publisher-independent global citation databases, and it allows users to explore the literature of various scientific fields.The literature retrieval details in this study are as follows: In the query, TI, PY, and DT stand for title, year published, and document type respectively.In order to improve the accuracy of literature retrieval, this study used search terms "HFpEF" or "heart failure with preserved ejection fraction" in the title to retrieve articles or reviews between 2009 and 2020.
Most of data analyses in this study were carried out with bibliometrix R-package (version 3.1), VOSviewer (version 1.6.17),[35] These bibliometrics tools can process the HFpEF literature collection, carry out various quantitative analyses, and build matrices for author collaboration, bibliographic coupling, references co-citation, and keyword co-occurrence networks.

Descriptive analysis
This study used WoS Results Analysis Tool and bibliometrix Rpackage to perform a descriptive analysis of the local HFpEF literature collection.In Table 1, some statistics information are reported.
The local HFpEF literature collection contained 1791 documents from 326 sources, including 1508 articles and 283 reviews.In these documents, a total of 7712 authors appeared 15,569 times and a total of 30,544 references were cited.WoS employs 2 types of keywords.Figure 1 shows the annual distribution of 1791 documents between 2009 and 2020.Except that the number of documents in 2019 decreased slightly compared with that in 2018, the number of documents showed an overall upward trend over year, increasing from 24 in 2009 to 353 in 2020, with an average annual growth rate of about 27.69%.
Figure 2 in which all the data were from WoS Results Analysis Tool, provides some snapshots of the top-10 relevant languages, research areas, countries/regions, and affiliations.
Figure 2A shows the relevant languages.English was the language most used by authors to write HFpEF documents.There were 1734 English documents, accounting for the vast majority of all documents.The other 2 languages used more were Russian and Spanish, with 34 and 13 documents respectively.In addition, a few documents were written in German, Portuguese, French, Polish, and Hungarian.HFpEF documents in the local HFpEF literature collection covered 43 research areas.Figure 2B shows top-10 most relevant research areas."Cardiovascular System Cardiology" was the leading research area having 1406 documents, followed by "General Internal Medicine," "Pharmacology Pharmacy," and "Radiology Nuclear Medicine Medical Imaging" with 143, 59, and 56 documents respectively.
A total of 79 countries/regions contributed to the global HFpEF document output.Figure 2C shows the top-10 most relevant countries/regions.Most were developed countries/ regions, of which USA ranked first with 807 documents, followed by Japan, Germany, and Netherlands with 202, 195, and 179 documents respectively.As the only developing country/

Author and source analysis
Authors and sources are important research constituents in bibliometrics.Figure 3 in which all the data were from R bibliometrix, used different criteria to rank authors and sources in the local HFpEF literature collection.
The top-10 authors in terms of the total numbers of documents are listed in Fig. 3A.All authors were having ≥50 documents, while Shah SJ ranked first, publishing 105 documents.Borlaug BA, Lam CSP, and Solomon SD were other major contributing authors with 89, 80, and 80 documents respectively.The most relevant authors ranking 5 to 10 were Redfield MM, Zile MR, Pieske B, Mcmurray JJV, Kitzman DW, and Anand IS.
The top-10 most productive sources are presented in Fig. 3B.They published a total of 707 documents which accounted for 39.5% of all documents included in the local HFpEF literature collection.The source published the most HFpEF documents was Eur J Heart Fail, with 133 documents, followed by Circ-Heart Fail, Am J Cardiol, and Int J Cardiol with 91, 76, and 72 documents respectively.JACC-Heart Fail, ESC Heart Fail, J Card Fail, J Am Coll Cardiol, J Am Heart Assoc, and Circulation were also sources that were clearly of interest to HFpEF research.
This paper introduced two concepts: global citation and local citation.Global citations means the citations that a local document has received from other documents indexed on WoS, while local citations means the citations that a reference, included in the bibliographies of the local documents, has Figure 3D lists the top-10 sources in terms of their local citations.Circulation topped the list with 9122 local citations.J Am Coll Cardiol ranked second, with a total of 8944 local citations.Eur Heart J was cited 4818 times locally, ranking third.
Author impact metrics can be used to evaluate the impact of the academic documents of an author or a source, and they are usually based on the number of citations and the number of documents.The commonly used author impact metrics include H-index, G-index, and M-index.H-index is the most used author impact metric, while G-index that gives more weight to highlycited document is an alternative to H-index. [36]This study adopted G-index to measure the local impact of authors and sources in the collection.The author local impact and source local impact by G-index are shown in Fig. 3E and F.
As shown in Fig. 3E, all authors showed G-index scores of ≥50, of which Borlaug BA had the highest G-index score of 88, followed by Shah SJ and Lam CSP.As shown in Fig. 3F, Eur J Heart Fail was the source with the highest G-index score of 77, and Circ-Heart Fail followed, with a G-index score of 76, one step away from the top of the list.All authors and sources in Fig. 3 played important roles on HFpEF research.
The top-10 authors with most local impact were identified by G-index. Figure 5 uses the timeline to show these top authors' production over the years.The production of an author referred to the number of published documents and the global citations of these documents.For example, Borlaug BA published 2 documents in 2009, which were cited 910 times in WoS since 2009.From Fig. 5, the research dynamics of these top authors can be observed.

Document and reference analysis
As listed in Table 2, top-20 most local cited documents included 17 articles and 3 reviews, of which "Paulus WJ, 2013, J Am Coll Cardiol" ranked first.All 20 documents were published before 2018.Further observation found that most of these documents    2, so there were duplicate items between the list in Table 3 and the list in Table 2.In order to compare the 2 lists, this paper added some special symbols to the duplicate items in Table 2, marked the corresponding special symbols in Note column in Table 3.New Engl J Med, J Am Coll Cardiol, Eur Heart J, and Circulation were the main sources of these most local cited references.After removing 6 keywords related to the search terms such as "heart failure," "heart failure with preserved ejection fraction," "preserved ejection fraction," "HFpEF," "left ventricular ejection fraction," and "ejection fraction," 169 keywords remained.
Burst keywords are regarded as indicators of research frontiers or emerging trends over time. [30]In this study, CiteSpace was used for burst keyword detection.A total of 54 burst keywords were found, and 20 meaningful keywords were selected and listed in Table 4. Keywords such as "hemodynamics," "comorbidity," "myocardial infarction," "inflammation," and "phenotype" have emerged in recent years.

Bibliographic networks
Bibliographic networks that include author collaboration network, bibliographic coupling network, references co-citation network, keyword co-occurrence network, and so on, can reveal the relationships between research constituents.This study used VOSviewer to perform bibliometric analysis and build networks.
Collaborations among authors have become more widespread as research methods and theories have become more complicated. [37]Contributions from a variety of authors can provide deeper insights, and their collaboration can raise the study level.Bibliographic coupling posits that 2 documents with public references have content that is similar. [37]Based on shared references, the analysis builds a similarity link between documents and classifies them into clusters.The clusters are constructed based on the citing documents, which is worth highlighting.In this analysis, 100 documents from the local HFpEF literature collection with the greatest global citations were chosen for bibliographic coupling analysis.Figure 9 and Table 5 show the details of the 4 clusters that were created from the 100 documents.
Another well-known bibliometric technique is co-citation analysis.This analysis assumes references that are cited together frequently are similar thematically. [37]When 2 references appear in a document's reference list at the same time, they are connected in a co-citation network.The most influential references and clusters can be discovered via co-citation analysis.It is worth noting that the clusters are derived based on the cited references.For reference co-citation analysis, 100 references most cited by the local documents were selected.Figure 10 and Table 5 demonstrate how the 100 references were divided into 3 clusters.
Keyword co-occurrence analysis is an approach that looks at the document's real content. [37]Keywords that commonly appear together are assumed to have a topic relationship in this analysis.Keyword co-occurrence analysis can reveal the study field's development trends and hotspots.From the 169 terms listed above, 100 keywords with the most occurrences were chosen for co-occurrence analysis.As shown in Fig. 11 and Table 5, the 100 keywords were grouped into 3 clusters.

Discussion
The pathogenesis of HFpEF is multifactorial. [27]Because of the lack of a complete understanding of pathophysiology of HFpEF, modern heart failure medication has not improved the prognosis of HFpEF patients.The treatment of HFpEF patients is still a challenge for clinicians, and further research is urgently needed to evaluate new therapies to improve the treatment effect and longterm prognosis of HFpEF patients.
Based on the literature of HFpEF research from 2009 to 2020, this paper made a bibliometric analysis to fully understand the research trend of HFpEF in the past 12 years.Annual scientific production can reflect the development trend of specific research fields.HFpEF research had been developing steadily during the period, which indicated that HFpEF research attracted more and more attention and was in the growth stage.
Nine of the top-10 countries/regions with the highest productivity are developed countries/regions, including 2 in North America, 6 in Europe, and 1 in Asia.Seven of the top-10 affiliations of global HFpEF document output are in USA, 2 in Europe, and 1 in Asia.In addition, most of top-authors worked in USA and Europe.Developed countries/regions, especially USA, were particularly influential in the field of HFpEF research, while the majority of Asian, African, and Latin American countries/ regions have relatively weak research capacity.There was an imbalance between countries/regions, and some research groups also lacked cooperation with the outside world.International cooperation that can pool the experience and wisdom of various countries/regions is an important way to accelerate scientific research progress and improve scientific research level.HFpEF research needs wider international cooperation.All sources involved in Fig. 3B, D, F and Tables 2 and 3, except that the JCR category of New Engl J Med, JAMA-J Am Med Assoc, and Lancet is "Medicine, General & Internal," the JCR category of other sources is "Cardiac & Cardiovascular Systems."As shown in Table 6, New Engl J Med, JAMA-J Am Med Assoc and Lancet are all top journals with high reputation.Most "Cardiac & Cardiovascular Systems" sources, such as Eur Heart J, Circulation, Eur J Heart Fail, etc, also have high influence.Most HFpEF related documents were published in "Cardiac & Cardiovascular Systems" sources, especially those on heart failure.There were also some HFpEF related documents published in "Medicine, General & Internal" sources such as New Engl J Med, JAMA-J Am Med Assoc, and Lancet.Documents published in sources with high influence would receive high attention, and their citations were often high.HFpEF related documents published in these sources need to be focused by researchers interested in HFpEF.
Compared to the reference Edlinger et al, [45] this work reveals new research hotspots in the field of HFpEF in recent years through keyword analysis, one of which is phenotype.Phenotypes refer to identifiable patient groups, which are distinguished by both the presence of predominant presenting symptoms and/or predominant comorbidity profiles. [23]The phenotypic heterogeneity of HFpEF is the main obstacle to HFpEF treatment. [22]Phenotypic classification of HFpEF patients based on comorbidities and treatable etiologies can help determine the optimal management strategy for each individual patient. [9]Therapeutics may need to shift from "one-size-fits-all" strategy to personalized approach based on phenotype characteristics. [19,22,38,39]After diagnosing HFpEF, a potential approach is to use biomarkers to identify HFpEF phenotypes, guiding towards more appropriate therapeutic strategies. [22]In addition, echocardiography can provide valu-  able information for further understanding of pathophysiology and potential phenotype of HFpEF. [39][42][43] Using machine learning to classify phenotypes of HFpEF patients without bias, high-quality evidence-based medical proofs may be produced, which can provide references for clinical decision-making.Oskouie et al [44] has developed a method of machine learning to analyze phenotypes, which divides HFpEF patients into 3 phenotype groups with different clinical characteristics and results.Segar et al [20] also used unsupervised clustering analysis method to classify HFpEF patients in high-dimensional mixed data queue.The results showed that cluster analysis based on machine learning can identify phenotype groups of HFpEF patients with different clinical characteristics.Using biomarkers from HFpEF patients for unsupervised cluster analysis, Woolley et al [25] identified 4 mutually exclusive patient subgroups according to the characteristics of biomarkers, which suggested different underlying pathophysiological pathways.In view of the phenotypic heterogeneity of HFpEF, treatment and clinical research need to be individualized and targeted.Researchers should abandon the mindset of "a pair of shoes for all feet," and use machine learning algorithm to divide HFpEF patients into different phenotype groups, which can make the patients better match the appropriate treatment and improve the prognosis. [21]Using machine learning to classify the phenotypes of HFpEF patients is a very noteworthy research hotspot.Despite an exhaustive bibliometric analysis of HFpEF study, deficiencies remain.Due to the time limit of WoS subscribed by our institution, this study could only query the literature after 2009, which might lead to the incompleteness of the local HFpEF literature collection.In addition, this study took title as the query item to search HFpEF literature, which improved the precision, but might reduce the recall.We have tried to search for HFpEFrelated papers using keywords, but the search results are not satisfactory.Many of the documents are not very relevant to HFpEF research and may only cover a small part of the content.We believe that HFpEF-related document retrieval by title will maximize the inclusion of the most important documents.For the annual analysis of HFpEF literature, the local HFpEF literature collection did not contain the latest HFpEF documents since 2021.In August 2021, however, an important breakthrough was reported.The sodium-glucose cotransporter 2 inhibitors were shown to be effective in the treatment of HFpEF. [46,47]This is another research hotspot worthy of attention.Furthermore, this study only processed the bibliographic database files from WoS, and did not consider other bibliographic databases, such as PubMed, Scopus, etc.These above deficiencies did not undermine the significance of this study.

Conclusion
With the sharp increase in the number of HFpEF patients, HFpEF is recognized as a major public health problem worldwide.Through bibliometric analysis of HFpEF literature from 2009 to 2020, this study provided a comprehensive reference and a summarized macroscopic overview of the research progress and performance related to HFpEF.
The bibliometric analytic tools employed in this study gleaned relevant information from published documents, supplemented standard review and survey articles, and allowed researchers to gain useful information.Over time, there has been an increase in the amount of HFpEF documents, however HFpEF research has been relatively imbalanced amongst countries/regions.It is necessary to conduct more cooperative and collective global HFpEF research in order to conduct HFpEF research in a balanced manner around the world.This study aids HFpEF researchers in locating possible collaborators and identifying research trends and hotspots, all of which will assist improve the treatment and prognosis of HFpEF patients around the world.
Data Source: Web of Science.Database: Web of Science Core Collection.Edition: Science Citation Index Expanded (SCIE).Search Option: Advanced Search.Query: TI = (HFpEF Or heart failure with preserved ejection fraction) and PY = (2009-2020) and DT = (Article Or Review).Record Content: full record and cited references.Data Format: Plaintext.Query Data: July 7, 2021.

Figure 3 .
Figure 3. Rankings of authors and sources.(A) Most relevant authors.(B) Most relevant sources.(C) Most local cited authors.(D) Most local cited sources.(E) Author local impact by G-index.(F) Source local impact by G-index.
received from the local documents.The diagram of global citations and local citations is shown in Fig. 4. In this paper, the concept of global citation is applicable to a local document, while the concept of local citation is applicable to a reference cited by the local documents.If a document has a high number of local citations, it means that it is an important document in the field.A local cited document is a local document cited by other local documents, so it belongs both in the document set and in the reference set.This paper lists the top-10 authors according to their local citations in Fig. 3C.The top-3 most local cited authors were as follows: Borlaug BA (3111 local citations), Redfield MM (2782 local citations), Shah SJ (2120 local citations).

Figure 4 .
Figure 4. Diagram of global citations and local citations.

Figure 5 .
Figure 5. Top-authors' production over the years.
were published between 2011 and 2013.This phenomenon indicates that the research foundation of HFpEF was basically formed between 2011 and 2013.Circulation followed by J Am Coll Cardiol were major contributors to these most local cited documents.Figure6uses the historical direct citation network to reveal the reference relationship between these top-20 most local cited documents.A total of 30,544 references were cited by the local documents.Top-20 most local cited references are listed in Table 3, and "Owan TE, 2006, New Engl J Med" topped the list.Among the 20 references, 11 were published during 2009 to 2020, of which 9 were included in the local HFpEF literature collection, and the other 2 were guidelines for the diagnosis and treatment of heart failure.The 9 local references also appeared in Table

Figure 6 .
Figure 6.Historical direct citation network of top-20 most local cited documents.
and Outcome of Heart Failure With Preserved Ejection Fraction Owan TE, 2006, New Engl J Med, V355, P251 Article 660 2 2016 ESC Guidelines for The Diagnosis and Treatment of Acute and Chronic Heart Failure Ponikowski P, 2016, Eur Heart J, V37, P2129 Review 459 3 A Novel Paradigm for Heart Failure With Preserved Ejection Fraction: Comorbidities Drive Myocardial Dysfunction and Remodeling Through Coronary Microvascular Endothelial Inflammation Paulus WJ, 2013, J Am Coll Cardiol, V62, P263 Review 440 4 How to Diagnose Diastolic Heart Failure: A Consensus Statement on the Diagnosis of Heart Failure with Normal Left Ventricular Ejection Fraction by the Heart Failure and Echocardiography Associations of the European Society of Cardiology 3.4.Keyword analysis Keywords are terms that reflect the concept of a document topic and can provide a reasonable description of the research hotspot.This study integrated Author Keywords and Keywords Plus for overall keyword analysis.The local HFpEF literature collection contained 4127 keywords (including Author's Keywords and Keywords Plus), and 3677 were left after manual processing of duplicate removal.The occurrences of 175 keywords reached or exceeded 20.

Figure 8 .
Figure 8. Author collaboration network.The color of a node is determined by the cluster to which the node belongs, and the size of the label and the circle of a node is proportional to the number of links of the node with other nodes.

Figure 9 .
Figure 9. Bibliographic coupling network.The color of a node is determined by the cluster to which the node belongs, and the size of the label and the circle of a node is proportional to the number of links of the node with other nodes.

Figure 10 .
Figure 10.References co-citation network.The color of a node is determined by the cluster to which the node belongs, and the size of the label and the circle of a node is proportional to the number of links of the node with other nodes.

Figure 11 .
Figure 11.Keyword co-occurrence network.The color of a node is determined by the cluster to which the node belongs, and the size of the label and the circle of a node is proportional to the number of links of the node with other nodes.

Table 1
Some statistical information about the HFpEF local literature collection.

Table 2
Top-20 most local cited documents.

Table 3
Top-20 most local cited references.

Table 4
Some burst keywords in the HFpEF research field.
HFpEF = heart failure with preserved ejection fraction.

Table 5
Descriptive information about bibliographic networks.

Table 6
Some sources involved in HFpEF research field.