Research Trends on High-Flow Nasal Cannula Oxygen Therapy: A Bibliometric Analysis from 2000 to 2019

Purpose: High-ow nasal cannula oxygen therapy (HFNC) has been increasingly reported over the past decades. Therefore, this study aimed to analyze the evolution of HFNC research and qualitatively and quantitatively evaluate publications. Methods: Publications between 2000 and 2019 were retrieved from the Web of Science Core Collection on August 17, 2020, according to a search strategy. Different kinds of software, CiteSpace and VOSviewer, and an online bibliometric analysis platform were used to identify top authors, journals, institutions, countries, keywords, co-cited articles, and trends. Results: Finally, 835 publications were identied, and the annual number of publications showed an increasing trend. Of these, 273 journals published articles related to HFNC. The United States and the University of Melbourne were the leading country and institution for publications, respectively. Co-cited reference analysis revealed top landmark articles. High-ow nasal cannula, bronchiolitis, hypercapnia, newborn, and immunosuppression are some of the high-frequency keywords in the co-occurrence cluster and co-cited reference cluster analyses, indicating that the clinical application of oxygen therapy remains a hot spot. Burst detection analysis of top keywords showed that randomized controlled trial, viral bronchiolitis, and immunocompromised patient were the new research foci. Conclusions: Bronchopulmonary dysplasia, infant, premature infant, respiratory distress syndrome, critical care patient, preoxygenation, extubation failure, and acute lung injury are the research foci of HFNC over the past 20 years and have emerged as a basis for transformation from infancy to adulthood and from a single disease to diverse diseases. high-ow nasal cannula, bronchiolitis, hypercapnia, newborn, immunosuppression, co-occurrence cluster patient keywords, using shows bronchopulmonary dysplasia, premature infant, respiratory distress critical care preoxygenation, extubation failure, acute lung randomized controlled


Background
The commonly used methods for oxygen inhalation include nasal catheter and face mask. Nasal catheter oxygen inhalation has advantages of simplicity, economy, convenience, and comfort, but when the oxygen ow is > 6 L/min, the dry and cold gas that does not reach ideal humidi cation will result in discomfort for patients, including frontal sinus pain, nasal mucosa dryness, and even bleeding [1]. The ow restriction directly affects the patient's inhaled oxygen concentration (FiO 2 ). Although mask oxygen inhalation can provide patients with higher FiO 2 , it will cause inconvenience while eating and drinking water, and patients will have a sense of restraint and occlusion, which are intolerable for many patients [2,3].
The high-ow nasal cannula oxygen therapy (HFNC) device can output high-ow gas with a constant oxygen concentration of 21-100%, temperature of approximately 37 °C, and relative humidity of 100%, thereby taking into account the advantages of ordinary nasal catheter and face mask. Moreover, as the output ow can reach 60 L/min, it produces a certain positive pressure and increases the functional residual volume [4,5], gradually making it another choice in addition to noninvasive ventilation (NIV) in adult patients with mild to moderate acute respiratory failure (ARF) [6][7][8], and plays an increasingly important role in patients with cardiac insu ciency [9,10] or patients undergoing respiratory tract invasive surgeries [11,12]. As an alternative to nasal continuous positive airway pressure, HFNC is widely used in neonates with acute respiratory distress syndrome [13], and its application in adult patients is increasing over time.
Bibliometric analysis is a statistical method based on a public literature database (such as the Web of Science) used to analyze and visualize a research trend [14,15]. Bibliometric analysis provides not only quantitative and qualitative evaluation of publications but also developing trends of a research domain. It can also present the most in uential research quickly and accurately, which provides a theoretical basis for further research. In addition, this information may be used as a policy guideline for decision makers [16]. Therefore, in this study, a bibliometric analysis of publications on HFNC published from 2000 to 2019 was performed. On the basis of our results, we will provide an overview of the achievements and future research trends and hotpots in this research domain.

Country/region And Institution Analysis
The 835 articles on HFNC research were published by research groups in 56 countries/regions. The top 10 countries (5 European countries, 2 Oceanian countries, 2 North American countries, and 1 Asian countries) published 794 articles, accounting for 95.09% of the total number of publications (Fig. 1B), with the United States as the leading country, which comprised 28.98% (242/835), followed by Australia and France, which accounted for 12.57% and 12.46% of the total number of publications, respectively. There were 1470 research institutions that published articles related to HFNC. The leading research institution with the highest number of publications was the University of Melbourne (58 articles), followed by the University of Toronto (53 articles), the University of Queensland (53 articles), and INSERM (41 articles). The research network among institutions showed a low-density map (density, 0.0251; Fig. 1C), indicating that the research groups are comparatively dispersed throughout various institutions, and thus, further academic collaborations are needed.

Journal Analysis
The total number of journals that published these 835 articles on HFNC was 273. The characteristics of the 10 most active journals are shown in Table 1. Most of the journal publishers are located in the United States and the United Kingdom. Respiratory Care published the highest number of articles on HFNC, followed by Pediatric Pulmonology and Intensive Care Medicine. As to the impact factor, Intensive Care Medicine has the highest, followed by Critical Care and Annals of Intensive Care. Regarding the JCR quartile, Intensive Care Medicine and Critical Care were ranked as Q1, BMJ Open and ACTA Paediatrica were ranked as Q2, Journal of Critical Care was ranked as Q3, and the ranks of Pediatric Pulmonology, Pediatric Critical Care Medicine, and Journal of Perinatology were dependent on the JCR partition.

Top Authors And Co-cited Author Analysis
Among all 3,788 writers in HFNC research, the top 10 most productive authors are shown in Table 2  VOSviewer keyword analysis of 835 articles identi ed 165 keywords with a minimum of 10 occurrences, and we divided them into ve clusters (therapy, noninvasive ventilation, positive airway pressure, high-ow nasal cannula, and intubation) ( Fig. 2A, 2B).

Top Co-cited Articles And Co-cited Reference Cluster Analysis
The co-citation and clustered network map were generated from 14,976 references using CiteSpace in a hierarchical order (Fig. 2C, 2D). Visualization of cocited articles showed a total of 529 nodes and 1,459 links (Fig. 2C). In this network, each node represents a cited article, and the size of each node is proportional to the total co-citation frequency of the associated article. As shown in Fig. 2D, co-cited references were clustered into nine major cluster labels, including oxygen therapy, vapotherm, bronchiolitis, newborn, sepsis bundles, immunosuppression, submicrometer aerosols, high-ow nasal cannula oxygen, and hypercapnia. Figure 2E displays a timeline view of distinct co-citation, showing that cluster #0 oxygen therapy and cluster #2 bronchiolitis had the highest concentration of nodes with citation bursts, and research foci seem to have shifted from vapotherm to oxygen therapy and bronchiolitis. This supports the nding of the emerging focus in HFNC. The top 10 co-cited articles and their cited frequency are shown in Table 3

Discussion
In this study, information visualization was used to analyze original articles on HFNC published from 2000 to 2019. On the basis of the identi ed trends that showed an increasing number of scienti c research publications over the 20-year period, we conclude that HFNC has become an interesting subject and an increasingly important area of research. Research themes on HFNC were relatively disorganized. Although the quantity of research was relatively considerable, integral analysis of research hot spots was lacking. Our intent was to catalog the attributes of relevant studies and focus on the interpretation of keyword cooccurrence and burst detection to predict and direct future research trends.
The conventional oxygen delivery methods have been nasal prongs, nose mask, and face mask, but the oxygen provided by conventional systems may not su ce in patients with certain diseases. HFNC systems utilize higher gas ow rates than the standard nasal cannula. The use of HFNC as a respiratory support modality is increasing in infant, pediatric, and adult populations as an alternative to noninvasive positive pressure ventilation [17][18][19][20][21][22][23][24]. High-ow oxygen (up to 60 L/min) via the nasal cannula, combined with a heated humidi cation system, may have several advantages. HFNC can decrease oxygen dilution, reduce respiratory dead space, and generate some positive airway pressure because of the expiratory resistance generated by the delivered continuous high ow [25]. Heated humidi cation can facilitate secretion clearance and decrease bronchial hyperresponse symptom development. HFNC is well tolerated and may be feasible in a subset of patients who require ventilatory support with noninvasive ventilation.
High-frequency keywords, including high-ow nasal cannula, bronchiolitis, hypercapnia, newborn, and immunosuppression, in co-occurrence cluster analysis and co-cited reference cluster analysis (Figs. 2A, 2B and 2D) indicate that the clinical application of oxygen therapy remained the hot spots in HFNC research, including different disease types and patient groups. Analysis of top keywords, using burst detection (Fig. 2F), shows that bronchopulmonary dysplasia, infant, premature infant, respiratory distress syndrome, critical care patient, preoxygenation, extubation failure, and acute lung injury attracted the most attention of peer researchers in the past 20 years, whereas randomized controlled trail, viral bronchiolitis, and immunocompromised patient were among the new research foci since 2016. Research foci in HFNC seem to have shifted from one disease to other diseases, and the patient group is no longer limited to infants or premature infants.
Journals with the highest number of articles in HFNC are mostly major journals related to critical care medicine and pediatrics, such as Critical Care, Intensive Care Medicine, Annals of Intensive Care, Pediatric Pulmonology, Pediatric Critical Care Medicine, and Journal of Perinatology. This indicates that HFNC has become a central topic of critical care and pediatric research. The analysis of the co-citation map of authors and top-cited authors (Fig. 1D, 1E and Table 3 Sztrymf et al. [31] enrolled 38 patients with ARF to evaluate the e ciency, safety, and outcome of HFNC. Reduced respiratory rate and increased pulse oximetry were observed as early as 15 min after the induction of HFNC, and PaO2 and PaO2/FiO2 signi cantly increased after 1 h HFNC as compared with baseline. No nosocomial pneumonia occurred during HFNC, and only nine patients required secondary invasive mechanical ventilation. A multicenter, randomized, unblinded trial on 830 patients showed that high-ow nasal oxygen therapy was not inferior to BiPAP (bilevel positive airway pressure) for patients with hypoxemia after a cardiothoracic surgery. Both methods had rapid effects on respiratory variables. BiPAP was associated with a higher PaO2:FiO2 ratio; high-ow nasal oxygen therapy, with lower PaCO2 values and respiratory rate. High-ow nasal oxygen therapy had no effect on the frequencies of adverse events or length of stay in the intensive care unit or hospital [33]. Frat et al. [6] compared the effects of high-ow oxygen through a nasal cannula, noninvasive ventilation, and standard oxygen through a face mask in patients with nonhypercapnic acute hypoxemic respiratory failure. The intubation rate at day 28 was lowest in the high-ow oxygen group compared with the other two groups, but did not reach a statistical difference. In addition, high-ow oxygen therapy, as compared with the standard oxygen therapy or noninvasive ventilation, resulted in reduced mortality in the ICU and at 90-day follow-up.
Patients with invasive mechanical ventilation usually use ordinary nasal catheters or face masks to inhale oxygen after extubation, whereas some patients with poor oxygenation often need to use an oxygen storage, Venturi masks, or even noninvasive positive pressure ventilation. Maggiore et al. [27] compared the effects of the Venturi mask and nasal high-ow (NHF) therapy on the Pa O2 /FI O2SET ratio after extubation and found that compared with the Venturi mask, NHF resulted in better oxygenation for the same set FI O2 after extubation. The use of NHF is associated with better comfort, fewer desaturations and interface displacements, and lower reintubation rate, suggesting the potential role of NHF in protecting extubation.
Chatila et al. [34] compared the effects of a high-ow oxygen (HFO) system to conventional low-ow oxygen (LFO) delivery at rest and during exercise in patients with COPD. They found out that delivering warm, humidi ed HFO improved exercise performance in a group of patients with severe COPD. An improvement in oxygenation with HFO at rest that was maintained during exercise despite similar-to-lower Fi O2 compared to LFO delivery was also observed.
More importantly, patients were less dyspneic and had lower arterial pressure even after performing longer exercises. A favorable change in the breathing pattern could also be associated with improved endurance during exercise while receiving HFO but not at rest.
Compared to traditional reviews, analysis based on CiteSpace provides a better insight of the evolving research foci and trends, but it comes with certain limitations. Similar words should be merged together during the analysis. Even though only original articles were included in the majority of the analysis, all article types were included during the co-cited reference analysis.

Conclusions
Our understanding of HFNC has undoubtedly signi cantly advanced via bursts of high-quality research occurring over the past 20 years. With the help of information visualization, research foci and overall trends in the eld were identi ed and provided information for future researchers. The more clinical application of HFNC would be the foci of future research.

Declarations
Competing interest The authors declare that they have no competing interests.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.