The Correlation between Altmetric Attention Score and Traditional Bibliometrics in Top Nursing Journal Articles

,


Introduction
Bibliometrics is the use of statistical methods to analyze publications, especially for scientifc contents [1,2].It is used to justify the role of the researchers and the research team [3] and also to identify key partners [4] and the progress of a specifc research feld [5].Te most widely used bibliometrics method is based on citations [6], such as citation times, impact factors [7], H index [8], Eigenfactor score [9], and CiteScore [9].However, these metrics have limitations.It would take time to accumulate citation times [10], and diferent journal articles could not be compared directly even in the same research feld [11,12].Absolute citation times favor older journal articles and have limited utility in comparing journal articles from diferent felds [12,13].Tus, new metrics have been developed.RCR [14], developed by NIH, is defned as the total number of citations that a paper received per year divided by the average feldspecifc citation rate for a peer companion group [14,15].It allows the comparison of articles in diferent felds.CNCI is another metrics to assess the paper's impact and has been analyzed in diferent studies [16,17], which can also be used to assess scientifc articles from diferent research felds [16,18].
In recent years, with the development of digital technology and the use of online platforms to discuss research, alternative-level metrics (altmetrics) have been introduced in research felds [12][13][14].Tey were used to measure the journal articles with attention, dissemination, overall infuence, and impacts [14].Te main altmetrics platform is Altmetric [19,20], which compiles the number of mentions of a paper across the most used social media platforms such as Twitter, Facebook, and LinkedIn, and public policy documents, mainstream media, online reference managers, and other online platforms, to generate a weighted score, Altmetric Attention Score (AAS) [14].AAS is a dynamic bibliometric which captures the online impact of a paper.It has been used to assess the online impact of journal articles across diferent felds [7,14,21] and provide potential evidence for research impact or journal strategy.
Nursing is an important discipline, and research helps the progress of the discipline by improving nursing practice and fnally helping nursing management.Factors that infuence the characteristics of nursing journal articles could further help journal articles to be quickly transformed into nursing practice and applied to nursing management.It has been reported that bibliometric analysis might help nursing management and nursing practice, and thus some studies have been performed in the feld of nursing management [2,16].However, these studies did not analyze the important impact of online platforms on nursing management.It is still unclear whether the online impact would be associated with traditional bibliometrics and whether the online impact would help nursing paper citations.Furthermore, it is also unclear whether the nursing manager could use social media, public research platforms, or online platfroms to help nursing research impact.Although several articles have used bibliometrics methods to assess the top cited [2,16] or the top impact nursing articles [2,16], whether impact from social media, public research platforms, or online platforms will help higher citations is a mystery.Terefore, in the current study, we analyzed the Altmetric and bibliometric data of journal articles published in nursing journals, to assess the correlation between AASs and the citation metrics.

Inclusion and Exclusion Criteria for the Selection of
Nursing Journal Articles.Te inclusion criteria were as follows: (1) the journal articles should be published in top nursing journals.Five-year impact factors were used to identify the top nursing journals, and thus the top 20

Paper Searching.
Te journal articles were searched by the selected journal names in the three databases (Incites [16], iCite [14], and Altmetric [12]).All search results were downloaded.If any data for an article was missing, the article was further searched in PubMed or the specifc journal websites.

Paper Screening and Data Extraction. All articles were screened according to the inclusion and exclusion criteria.
Only articles which had all the required data (RCR, citation time, AAS, and CNCI) were included.Te following information was extracted from each article: journal, author, impact factor, RCR score, citation time, CNCI value, AAS, etc. Te RCR scores were extracted from the downloaded fle from the iCite database, the CNCI values were extracted from the downloaded fle from the Incites database, and the AASs were extracted from the downloaded fle from the Altmetric database.Te journals' impact facotrs were downloaded from the Web of Science.

Statistical
Analysis.Te characteristics of the included studies were analyzed.Te correlations between AASs and citations, AASs and RCRs, AASs and CNCIs, and AASs and impact factors, were analyzed using Spearman correlation coefcients [12].Subgroup analyses were performed according to journals, and years.All analyses were performed by SPSS 25.0 software.

Te Main Characteristics of Included Journal Articles.
Te data were collected by September 9 th , 2022.After screening all data of the journal articles according to the inclusion and exclusion criteria, a total of 15,212 journal articles were included.Te impact factors for all the selected 20 journals from 2010 to 2019 are shown in Table 1, and the impact factors of most journals increased from 2010 to 2019.Te average citation of all journal articles was 17.36 times, and the median citation was 11 times.A total of 318 journal articles were cited 0 times, 14,059 journal articles were cited 1 to 50 times, 660 were cited 51 to 100 times, and 175 journal articles were cited more than 100 times.

Te Performance of Included Journal
Articles.Te average RCR score of included articles was 1.71, and the median RCR score was 1.14.Te RCR scores of 258 articles were 0; 6,576 articles had RCR scores between 0 and 1; 8,236 journal articles had RCR scores between 1 and 10; the RCR scores of 142 journal articles were higher than 10.Te average CNCI value was 1.52, and the median CNCI value was 1.05.Te CNCI values of 318 journal articles were 0; 7,017 journal articles had CNCI values between 0 and 1; 7,785 journal articles had CNCI values between 1 and 10; the CNCI values of 92 journal articles were higher than 10.Te average AAS was 8.92, and the median AAS was 3. Te AASs of 2264 journal articles were 0; 9,950 journal articles had AASs between 0 and 10; 2,866 journal articles had AASs between 10 and 100; the AASs of 132 journal articles were higher than 100.Te citations, RCRs, CNCIs, and AASs of all journal articles in the 20 journals were summarized in listed in Table 2.

Te Contribution of Diferent Sources to AAS.
Te sources that contributing to the AAS were individually analyzed from 2010 to 2019.For all years, Twitter was the most frequent contributor to AAS followed by news outlets and Policy (Table 3).
For the International Journal of Nursing Studies, which ranked frst in the JCR report (2021) with an impact factor of 6.612, a total of 1,437 journal articles were included.Te total citation times was 43,729 (ranging from 0 to 582 by paper, median 19), the total AAS was 12,975 (ranging from 0 to 703 by paper, median 3), and the total RCR score was 3,985.92(ranging from 0 to 35.94 by paper, median 1.79).Very weak correlations were found: 0.207 (95% CI, 0.158-0.258)between AASs and citations, 0.371 (95% CI, 0.325-0.418)between AASs and RCRs, and 0.267 (95% CI, 0.219-0.317)between AASs and CNCIs.
Te Journal of Clinical Nursing ranked fourth in the JCR report (2021) with an impact factor of 4.423, from which we collected the highest number of journal articles in the current study.Te total number of citations was 48,236 (ranging from 0 to 262 by paper, median 10), the total AASs was 17,842 (ranging from 0 to 251 by paper, median 2), and the total RCR score was 4,709.72 (ranging from 0 to 21.32 by paper, median 0.98).Very weak correlations were found: 0.133 (95% CI, 0.097-0.166)between AASs and citations, 0.312 (95% CI, 0.28-0.346)between AASs and RCRs, and 0.23 (95% CI, 0.199-0.264)between AASs and CNCIs.
Te Journal of Nursing Management was the only management journal in the feld of nursing, from which a total of 783 journal articles were included.Te total number of citations was 13,906 (ranging from 0 to 203 by paper, median 11), the total AASs was 3,007 (ranging from 0 to 228 by paper, median 1), and the total RCR score was 1,593.84(ranging from 0 to 17.58 by paper, median 1.43).Very weak correlations were found: 0.158 (95% CI, 0.086-0.229)between AASs and citations, 0.239 (95% CI, 0.162-0.304)between AASs and RCRs, and 0.206 (95% CI, 0.137-0.274)between AASs and CNCIs.
Nurse Education Today was the only education journal in the feld of nursing, from which a total of 1,566 journal articles were analyzed.Te total citation was 26,555 (ranging from 0 to 770 by paper, median 11), the total AAS was 15,837 (ranging from 0 to 373, median 3), and the total RCR score

Te Correlation Analysis by
Years.Spearman rank correlation coefcient was calculated between citations and the AASs by years (Table 5).Te results showed very weak correlations between AASs and citations, AASs and CNCIs, and AASs and RCRs in all years.

Te Correlation between AASs and Impact Factors.
Te correlations between AASs and impact factors were individually analyzed based on years.Weak correlations were found in most of the years (Table 6).Spearman rank correlation coefcient was calculated between the impact factors and AASs based on journals (Table 7).Weak correlations were found for the Australian Critical Care, International Journal of Mental Health Nursing, International Journal of Nursing Studies, Journal of Advanced Nursing, and Nurse Education Today.

Discussion
In the current study, the Altmetric and traditional bibliometric of journal articles published in 20 top nursing journals were analyzed.A very weak correlation between AASs and citations was found.AAS was also found to be correlated with RCR and CNCI, which were novel metrics of research infuence based on citation times.Te results suggest that traditional bibliometrics and AAS cannot be used interchangeably but rather complementarily when assessing the impact of journal articles in the nursing feld.
Further analysis by journals and publication years found that the International Journal of Nursing Studies (IJNS) had more correlation between the AASs and citations, and no correlation was found in the Nursing Ethics.Tis diference might be due to the diference in the breadth of journal topics since one is a comprehensive nursing journal and the other is a subspecifcity nursing journal.Articles from IJNS with high online attention might also impact the scientifc community and thus increase citations.However, articles from Nursing Ethics, which are likely to be controversial articles for the public to garner public interest, thereby increasing AAS, while they might not impact the scientifc community in the same manner [12].
With the COVID-19 pandemic [22], more training programs in the nursing practice feld have been performed via the social media platforms and other online platforms.Tat should be of interest to nursing managers, nurses, and journals.It indicates that the social media and other online platforms would not only help the scientifc community but also help the public community, especially during the COVID-19 pandemic period [23].Implementing new research works on the online platforms will contribute to these results and further help the scientifc community.
Te study has several strengths.First, the current study was a study with large sample size analyzing alternative metrics and bibliometrics based on articles in top nursing journals.Second, it analyzed the AASs with 3 citation-based metrics and found weak correlations between these parameters.Tird, we analyzed the correlation by journals, which was a unique contribution to the nursing feld.Te paper also has several limitations.First, to analyze the precision correlations, we excluded journal articles that missed any data of the four variables (AASs, citations, RCRs, and CNCIs).Second, the correlations might be diferent when analyzing at a diferent time.Tird, the study only   analyzed the journal articles from top nursing journals, and nursing-associated journal articles published in other journals could have diferent kinds of correlations, especially for those articles published in low-impact factor journals.
Fourth, the study did not include other bibliometrics [24,25], such as the Eigenfactor score and CiteScore; in the future, a new study should be performed to assess these correlations.
In conclusion, the study of articles published in highimpact nursing journals between 2010 to 2019, fnds very weak correlations between AAS and citation-based metrics, which suggests promoting journal articles on online platforms may help research, journals, and nurses.Future studies are needed to assess the long-term correlations among these metrics for nursing journal articles.

Implication to Nursing Management
Understanding the correlation of online impact with traditional bibliometrics of research is critical to nursing practitioners, which in turn helps manage nursing journal articles and research.Nursing managers should develop targeted strategies to increase the online impact of research or nursing practice and increase research impact.

Table 1 :
Te impact factors of the included 20 nursing journals from 2010 to 2019.

Table 3 :
Attention received by papers from high-impact journals from 2010 to 2019 stratifed by diferent resources.

Table 4 :
Te correlations between citation and AAS, CNCI and AAS, RCR, and AAS based on journals.