‘Compare’ Google scholar & PubMed for Drug Searches

If ever a topic cries out for the participation (and input) of health librarians, it’s a recently-published comparative study of Google scholar and PubMed in retrieving the drug literature entitled “Google scholar versus PubMed in locating primary literature to answer drug-related questions“. The paper – published in this month ‘s Annals of Pharmacotherapy – is authored by three Pharm Ds and one PhD in Business at Samford University, Alabama.

A brief look at the structured abstract suggests the authors have something important to say about the two popular medical search tools. They cite much of the seminal literature in using Google scholar in medicine, which is commendable. Their objective? To compare Google scholar and PubMed as to the coverage of primary drug literature, # of citations, access to “free fulltext articles” – within the first 100 results.

My problem here is I don’t believe the authors acknowledge the true dissimilarities of Google scholar and PubMed. The databases are, for all of their overlap, apples and oranges. The authors get the search limitations of each database but some subtleties are lost. For example, PubMed is constructed (and searches) very differently than Google scholar; PubMed is a proper medical database & search tool whereas Google scholar is a browsing tool.

The authors should talk to their local health librarians to discern the aspects/concerns/issues of searching these two tools. By comparing the first “100 results” in each database, they touch that PubMed lists articles by when items are received and/or indexed whereas Google scholar lists articles based on link love (and a secret Google algorithm). However, straight counts in GS cannot be done because we don’t know what the database contains, what authority work is done and which journals it indexes.

But can this really be news? Further, they should consider many of Jacso’s critiques of Google scholar as his writing is certainly central to the GS debate. Tut, tut Annals of Pharmacotherapy!


1. UBC Health Library wiki. Google scholar bibliography http://hlwiki.slais.ubc.ca/index.php/Google_scholar_bibliography

6 thoughts on “‘Compare’ Google scholar & PubMed for Drug Searches

  1. I will leave it to my colleagues to address the bulk of the above comments, should they wish (though it appears to me that the thrust of the article alluded the reviewer)…

    As for the remark that the paper was “authored by three Pharm Ds and one PhD in Business [snip]” I can only speak for myself, the “PhD in Business.” The kindest thing that I can say is that the statement was presumptiously ignorant; I am a statistician; having spent a significant portion of my career as a research biostatistician in academic medical centers prior to joining the faculty of the Brock School of Business (see http://faculty.samford.edu/~twwoolle/CV.htm) leads me to believe that I am more than qualified to have collaborated on the referenced study. By the way, the “three Pharm Ds” are top-notch scholars with whom I am honored to be associated.

  2. Hi Thomas
    Thank you for your message. I have muted some of my comments on the blog and was going to email you specifically to apologize. I think I came across as harsh whereas my intention was to start some dialogue.

    I think that a librarian might have helped you in aspects of the study. I was also pleased to learn more about your background as a statistician. I should do my homework next time!~

  3. Dean, thank you for blogging about this. I am a former medical librarian with a PhD in medical informatics/library and information science, full-time faculty at a library school in the Midwest which you can find by investigating my website. I teach courses in consumer health information, health information systems, and online searching, to audiences of upper-level undergraduates through doctoral students in LIS and engineering. I build into all these courses, but especially the online searching one, comparisons of search results from Google and Google Scholar with search results from commercial, structured, focused databases, including Web of Knowledge products but also PubMed databases.

    Dean is quite right when he says “However, straight counts in GS cannot be done because we don’t know what the database contains, what authority work is done and which journals it indexes.” I find that most of my students don’t understand this; they believe the entire universe of scholarly literature is being indexed by Google. They base their assumptions of Google content on their knowledge about commercial database content, which is ironic under the circumstances. And they see anything Google provides them as deriving from a search of a *collection*.

    But a collection of what? The fact is that without knowing who Google’s partners are, we don’t know what kind of “collection” it is, which IMHO severely limits its usefulness as a single source. I like the study — I think it was well done — and I plan on pointing my students to it in a few weeks when we examine Google Scholar in depth. But I don’t think it’s possible for this or any other study to be any more than a snapshot of Google Scholar’s capabilities at a single point in time. Analyzing Google Scholar’s output as if it were analogous to PubMed’s output would make sense if they were analogous bodies of literature in the first place — but they aren’t.

  4. I share a lot of Catherine’s thoughts – of course, because I’m also a medical librarian. I think the question was good, if only because we have so many people that put Google Scholar in high regard. As I tell my physicians, residents and students – it is a great tool, but it isn’t great all the time. You need to know when to use which tool for which purpose. A driveway can be cleared of snow with a spoon, but why not use a shovel?

    Additionally, I found it interesting that a medical librarian wasn’t consulted, or if so not acknowledged. I think, and I may be alone here, that a few parts of the study could have been avoided simply by talking to a librarian. For example, on the third page of the article there’s discussion about determining how results are returned. There’s no question about how results are displayed in PubMed – by default the system lists them chronologically, newest first. Yes, it will return searches other ways, but has to be told to do so, and the authors make no reference of changing settings.

    I too have referred people to see the article because there are some valid findings, but again, I’m concerned that a librarian wasn’t consulted. I’m not sure I’d do drug research and not think consult an expert in pharmacology.

  5. Having done online searching frequently for several corporations in the biopharmaceutical industry since 1983 and with the 7 years previous to that working as an Editor/Indexer for BIOSIS, I would like to think I know a bit about indexing, thesaurus, biomedical literature searching, and database selection.

    I think we are all missing the point. I feel the most important conclusion of this article was unfortunately buried and somewhat lost in the Limitations section:

    “…searchers need to explore the strengths and weaknesses of both databases and should not regard Google Scholar as the definitive search engine for published medical literature…”

    To this, I would like change “both databases” into “both databases plus other fee-based systems such as EMBASE.com” and also change “Google Scholar” to “Google Scholar and/or PubMed”.

    Thus, anyone doing searches in PuMed or Google Scholar using drug trade names will be severely disappointed.

    You get what you pay for. Free sites cannot be expected to yield highly comprehensive or highly accurate search results, regardless of the expertise or the qualification of the searcher.

    Only an experienced online searcher with appropriate library background AND subject matter expertise who can judiciously select which databases to search in, what to include in the search strategy, which thesaurus/taxonomy to deploy, is capable of producing the most precise search result.

Leave a Reply

Your email address will not be published. Required fields are marked *