Comparing Bibliometric Analysis Using PubMed, Scopus, and Web of Science Databases

Summary

Literature databases are commonly used to assess publications in a certain subject, discipline, country, or region of the world, a practice known as bibliometric analysis. The current protocol details how to use PubMed, Scopus, and Web of Science databases to do bibliometric analysis.

Abstract

Literature databases (i.e., PubMed, Scopus, and Web of Science) differ in terms of their coverage, focus, and the tool they provide. PubMed focuses mainly on life sciences and biomedical disciplines, whereas Scopus and Web of Science are multidisciplinary. The protocol described in the current study was used to search for publications from Jordanian authors in the years 2013-2017. In this protocol, how to use each database to conduct this type of search is explained in detail. A Scopus search resulted in the highest number of documents (11,444 documents), followed by a Web of Science search (10,943 documents). PubMed resulted in a smaller number of documents due to its narrower scope and coverage (4,363 documents). The results also show a yearly trend in: (1) the number of publications, (2) the disciplines that have the most publications, (3) the countries of collaboration, and (4) the number of open access publications. In contrast, PubMed has a sophisticated keyword optimization service (i.e., Medical Subject Heading, or MeSH), while both Scopus and Web of Science provide search analysis tools that can produce representative figures. Finally, the features of each database are explained in detail and several indices that can be extracted using the search results are provided. This study provides a base for using literature databases for bibliometric analysis.

Introduction

Classically, researchers have used literature databases to perform literature review for their studies¹. Another use of these literature databases arose at the end of the 19^th century, where researchers analyzed the body of literature, a use that has slowly grown since². In the last few decades, digitizing literature and the formation of online literature databases provided an opportunity to researchers to analyze the body of literature and research performance easily and efficiently. An example would be analyzing the research performance for a document³, a subject⁴, a discipline⁵, a country⁶, or even a region in the world⁷. This type of analysis is known as bibliometric analysis. Heartsill Young defined bibliometric analysis as the use of statistical methods to analyze a body of literature to reveal historical development⁸. In other words, bibliometrics is the quantitative study of published units on the basis of citation and text analysis⁹.

Different databases are used to do bibliometric analysis and each database has different characteristics and can provide different services¹⁰. Currently, the most commonly used literature databases are the Web of Science and Scopus for almost all disciplines, both only available on a subscription basis¹¹, and PubMed for biomedical and life sciences, a freely available database¹⁰. There is also Google Scholar, which might be an easy tool to handle, but it should not be used as a bibliometric analysis tool currently due to some deficiencies such as its unclear scope and coverage, its lack of citation analysis tools, and its inclusion of non-peer reviewed non-scientific contents^12,13. Moreover, Google Scholar lacks the tools for performing advanced search and keyword optimization¹⁴.

Several previous studies have compared the features of the previously mentioned literature databases for literature review purposes^{3,5,10,12,13,15,16,17}. However, in this study, the means by which PubMed, Scopus, and Web of Science databases are used to perform a bibliometric analysis will be provided, and the pros and cons for each of them will be compared. Bibliometric analysis can be used to analyze the research output in almost any discipline, so the target audience would be any researcher who intends to analyze publication trends. An example of analyzing a publication trend in Jordan as a country will be presented using each database. Jordan was chosen because doing a bibliometric analysis for a country (in contrast to a subject) is not very straightforward. In addition, Jordan, specifically, is poorly studied in a bibliometric way as it can be both an author name and a country name. We explain how to overcome such a challenge in the search.

Protocol

NOTE: The following are search methods and an example search for each method is provided. Note that the part related specifically to bibliometric analysis is also supplied.

1. PubMed

1. Choose Advanced Search from PubMed homepage (www.ncbi.nlm.nih.gov/pubmed).
2. Enter the desired search term in the search field. Choose the search terms from the medical subject heading (MeSH) database. The following example details how to assess research in "cancer".
   1. Open the MeSH database: https://www.ncbi.nlm.nih.gov/mesh.
   2. Enter the word "cancer" in the search field.
   3. Ensure that the results show the word "neoplasms" at the top of the search, where it notes that using the word "neoplasms" is more suitable, as "neoplasms" is the word used to index relevant topics.
   4. Further check other terms listed under the term "neoplasms" by clicking on it. The results will show other terms used to describe similar topics (e.g., tumor), and will also list other subcategories (subheadings) under the term neoplasm.
   5. Use the drop-down lists in the Builder section to specify the field of the article that PubMed will search for the term in. Note that the following search fields are available: all fields, title, abstract, authors, affiliation, conflict of interest, language, journal, publisher, publication type, grant number, ISBN, and MeSH terms.
   6. Add as many fields as needed and choose the relation between these fields (AND, OR, or NOT). See Table 1 for further details.
3. Click Search.
4. Refine the search results further in the results using different filters available as detailed in Table 2. Note that from now on, the final search done will be saved in the history of the Advanced Search, which was accessed in the first step. This means that the search can be paused at this step and resumed later.
5. Ensure that each new search is saved in the Advanced Search history screen where it will be assigned a number (e.g., #2). Use this number in the search field above to subtract searched queries (e.g., "#1 NOT #2" to subtract results in search #2 from results of search #1).
6. Export the results to further analyze them.
   1. Use the FLink tool (https://www.ncbi.nlm.nih.gov/Structure/flink/flink.cgi) to export the results in a comma separated values (CSV) format, and choose PubMed from the Please choose a database to start drop-down list.
   2. Choose Input From Entrez History from the input screen, and an advanced PubMed search history will appear at the drop-down list.
   3. Choose the search performed in the previous steps at the PubMed advanced search and click Submit.
   4. View the resulting report since it provides the option of exporting the search results in a CSV format file.
7. Perform the following steps to analyze Jordanian research output during a 5-year period between 1/1/2013 and 31/12/2017 using PubMed.
   1. Open the Advanced search form in the document search form at the PubMed website (www.ncbi.nlm.nih.gov/pubmed). 
   2. Use Jordan as the search term and specify Affiliation as the search field. Note that PubMed interprets affiliation as all information related to the author (i.e., author's name, address, affiliation), so exclude any document authored by an author named "Jordan" in which the affiliation country is not Jordan. Follow the steps below to avoid including such irrelevant results.
   3. Type Jordan in another field and choose the field type Author.
   4. Choose the operator NOT as the relation between the two fields and click on Search.
   5. Specify Publication Dates from 1/1/2013 to 31/12/2017 in the results window, and choose Journal Article and Review from Article Types.
   6. Open FLink (https://www.ncbi.nlm.nih.gov/Structure/flink/flink.cgi), and choose PubMed as the database.
   7. Choose Input From Entrez History from the input screen, and choose the search from the drop-down list.
   8. Click Download CSV.
      NOTE: Figure 1 details the PubMed search report with annotation for each section of the report.

2. Scopus

1. Register in order to access the full search capabilities of the Scopus database. Check if the local institution is already registered and has access to the database since academic institutions are usually registered in Scopus.
2. Go to the website (www.scopus.com) and by default, Scopus opens the document search form screen.
3. Enter the search term desired in the search field available.
4. Specify the fields in the article to be searched for. Note that the following search fields are available: all fields, title, abstract, keywords, authors, affiliations, funding information, language, references, conference, ISSN, CODEN, DOI, ORCID, and CAS number.
5. Add other fields to search for and indicate the relation between the newly added field and the other field already entered (AND, OR, or AND NOT). See ( Table 1) for further details.
6. Use the Limit option to limit the search based on the options provided by Scopus, as detailed in Table 2.After executing the search, save the search and continue later if needed.
   1. In this case, set an alert using the option Set alert, where an email will be sent when an article satisfying the searching criteria is added.
7. Refine the search results further from the results directly by choosing from the options provided by Scopus ( Table 2), where Scopus shows the number of documents included for each option.
8. Choose to either analyze the results directly on the Scopus website (click Analyze Search Results), or to export the results in either zip or CSV formats after completing the search criteria.
9. Perform the following steps to analyze Jordanian research output during a 5-year period between 1/1/2013 and 31/12/2017 using Scopus.
   1. Go to the website (www.scopus.com) and by default, Scopus opens the document search form screen.
   2. Type Jordan as the searching term in the document search form.
   3. Specify Affiliation Country as the search field.
   4. Limit the search duration from 2013 to 2017. Note that from 2013 means from 1/1/2013, and to 2017 means to 31/12/2017.
   5. Limit the document type to Article or Review, and then click Search.
      NOTE: Figure 2 details the Scopus search report with annotation for each section of the report.

3. Web of Science

1. Register in order to access the full search capabilities of the Web of Science database. Check if the local institution is already registered and has access to the database as academic institutions are usually registered in Web of Science.
2. Go to the Web of Science home page (www.webofknowledge.com). The website opens the basic search and includes the Web of Science Core Collection as the selected database for search.
3. Search for the fields as detailed in Table 2.
4. Add another field (if needed) to connect both fields by either AND, OR, or NOT. See Table 1.
5. Define the duration that is searched down to 1945. After completing the search, the results are saved in the history and can be returned to at any time. If needed, set an alert if any new document is added to the search report.
6. Sort the results according to either data, times cited, usage counts, or to other categories from the drop-down list provided.
7. Refine the search results further from the results directly by choosing from the options provided by Web of Science ( Table 2), where Web of Science shows the number of documents included for each option.
8. View the results and analyze them via a tree map or bar graph. Note that there is a table showing the count in each category.
9. Download the results. Note that unlike Scopus and PubMed, Web of Science only allows the downloading of 5,000 records at a time (e.g., a 10,000 result search is downloaded in two batches, the first batch for the first 5,000 records, and a second batch for the next 5,000 records).
10. Perform the following steps to analyze Jordanian research output during a 5-year period between 1/1/2013 and 31/12/2017 using Web of Science.
    1. Type Jordan in the search field and specify Address as the search field.
    2. Identify the duration of search between 2013 and 2017 and click Search.
    3. Restrict the search using Article and Review filters.
    4. Choose to analyze the results that are saved in Search History now or later.
    5. Choose to analyze the results in the form of tables or visual tree map and bars.
       NOTE: Figure 3 details the Web of Science search report with annotation for each section of the report.

Results

Results from PubMed search

A total of 4,363 documents were retrieved based on the search conducted in this study. Free full text was available for 1,767 documents (40.5%). In 2013, a total of 532 documents were published, 663 documents in 2014, 811 documents in 2015, 952 documents in 2016, and 1,405 documents in 2017.

The results reveal that 1,008 (23....

Discussion

In this study, the steps through which PubMed, Scopus, and Web of Science databases are used to perform a bibliometric analysis were provided. It was indicated that the friendliest and the easiest tool to use for bibliometric analysis services is Web of Science; however, its drawback is that its services are not available for free. PubMed is devoted for biomedical sciences and is affiliated with several other National Library of Medicine (NLM) tools that can help to optimize analysis of biomedical subjects. Medical Subje...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
	clarivate	N/A	Web of Science provider, where the access was provided by the subscription made by the University of Jordan.
	Elsevier	N/A	Scopus provider, where the access was provided by the subscription made by the University of Jordan.