Google Scholar Versus PubMed in Locating Primary Literature to Answer Drug-Related Questions

Maisha Kelly Freeman, PharmD BCPS; Stacy A Lauderdale, PharmD BCPS; Michael G Kendrach, PharmD FASHP; Thomas W Woolley, PhD

Disclosures

The Annals of Pharmacotherapy. 2009;43(3):478-484. 

In This Article

Abstract and Introduction

Abstract

BACKGROUND: Google Scholar linked more visitors to biomedical journal Web sites than did PubMed after the database's initial release; however, its usefulness in locating primary literature articles is unknown.
OBJECTIVE: To assess in both databases the availability of primary literature target articles; total number of citations; availability of free, full-text journal articles; and number of primary literature target articles retrieved by year within the first 100 citations of the search results.
METHODS: Drug information question reviews published in The Annals of Pharmacotherapy Drug Information Rounds column served as targets to determine the retrieval ability of Google Scholar and PubMed searches. Reviews printed in this column from January 2006 to June 2007 were eligible for study inclusion. Articles were chosen if at least 2 key words of the printed article were included in the PubMed Medical Subject Heading (MeSH) database, and these terms were searched in both databases.
RESULTS: Twenty-two of 33 (67%) eligible Drug Information Rounds articles met the inclusion criteria. The median number of primary literature articles used in each of these articles was 6.5 (IQR 4.8, 8.3; mean ± SD 8 ± 5.4). No significant differences were found for the mean number of target primary literature articles located within the first 100 citations in Google Scholar and PubMed searches (5.1 ± 3.9 vs 5.3 ± 3.3; p = 0.868). Google Scholar searches located more total results than PubMed (2211.6 ± 3999.5 vs 44.2 ± 47.4; p = 0.019). The availability of free, full-text journal articles per Drug Information Rounds article was similar between the databases (1.8 ± 1.7 vs 2.3 ± 1.7; p = 0.325). More primary literature articles published prior to 2000 were located with Google Scholar searches compared with PubMed (62.8% vs 34.9%; p = 0.017); however, no statistically significant differences between the databases were observed for articles published after 2000 (66.4 vs 77.1; p = 0.074).
CONCLUSIONS: No significant differences were identified in the number of target primary literature articles located between databases. PubMed searches yielded fewer total citations than Google Scholar results; however, PubMed appears to be more specific than Google Scholar for locating relevant primary literature articles.

Introduction

Google promotes itself as the "world's best search engine" due to its speed, accuracy, and ease of use.[1] Healthcare professionals are using Google to search for a variety of topics, including the diagnosis of medical conditions. For example, a medical resident correctly diagnosed a rare disease by entering relevant search terms into Google.[2] In addition, investigators have also determined the percentage of correct diagnoses arrived at from Google searches for cases published in a case record of the New England Journal of Medicine. Google searches yielded a correct diagnosis for 15 of 26 cases (58%; 95% CI 38 to 77).[3]

Google Scholar is a beta-phase search engine produced by Google that was introduced in November 2004. Google Scholar searches retrieve results that include scholarly literature citations as well as other literature, including peer-reviewed publications, theses, books, abstracts, and other articles from academic publishers, professional organizations, and preprint repositories, universities, and other scholarly organizations.[1] Therefore, Google Scholar is able to retrieve more types of literature (ie, books, theses, abstracts) compared with medical literature database retrieval search engines like PubMed.

Literature in Google Scholar is ranked by the availability of full-text articles, author name, weight of the publication in which the article appears, and the number of times the article was previously cited. Articles that are most relevant to the search terms are promoted to be located on the first page of displayed results.[1] Since Google Scholar's release in November 2004, use has increased rapidly. More visitors were led to biomedical journal Web sites by Google Scholar within a year of release than by PubMed.[4]

PubMed is a free Internet database published by the National Library of Medicine that serves as an indexing service for biomedical literature.[5] PubMed indexes over 17 million citations in greater than 5000 biomedical journals. Due to the availability of free full-text articles and its search features, PubMed has become the most frequently used biomedical indexing database.[6]

Several investigators have determined the comparative number of citations (ie, total number of results retrieved by each database) indexed in Google Scholar versus other indexing databases.[7,8] Shultz[7] conducted a variety of test searches in PubMed and Google Scholar databases, based on questions received from researchers searching for literature citations or those that were previously developed for use during instruction for medical librarian searches, to determine whether the same number of citations could be retrieved between databases. Google Scholar retrieved more results than PubMed in 8 of 10 searches. However, it is unknown whether the results located by Google Scholar searches were from medical literature versus other types of literature. Conversely, the number of clinical practice guidelines that could be retrieved from Google Scholar compared with SumSearch (a unique method of searching for medical literature via the Internet that includes general search engines and MEDLINE[9]) was assessed.[8] Google Scholar retrieved fewer citations (34.9%) compared with SumSearch (65.1%). Due to conflicting results between the 2 studies, the proficiency of Google Scholar versus PubMed in locating medical literature is unknown.

To our knowledge, no published investigations have assessed the number of primary literature (ie, original medical research) citations that are retrieved from PubMed and Google Scholar searches if identical searches are conducted within both databases. Primary literature citations were targeted because the main focus of PubMed is to locate primary literature articles, information in primary literature is used to make clinical decisions, and there may not be an adequate number of published review articles that address unique clinical dilemmas or questions to aid in the determination of a clinically relevant response. The purpose of this study was to assess the usefulness of both databases for locating target primary literature to answer specific drug information questions by evaluating the availability of primary literature articles; total number of PubMed and Google Scholar citations retrieved; availability of free, full-text journal articles; and target number of primary literature articles located by year within the first 100 citations retrieved from these 2 databases.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....