Information Resources for Medical Students: FAQs and Handouts

The HSC Library has several databases that are collections of electronic reference books - these would be good to search for background information. Examples include Access Medicine, Access Pharmacy, and Stat!Ref.

Selecting the appropriate database really depends upon the search topic and the purpose of the search. PubMed is usually a good place to start. If you’re looking for a straight-forward clinical answer, a textbook database like Access Medicine or a point-of-care database like UpToDate may be sufficient.

EXAMPLES:
• For questions relating to health administration, public health, or allied health, you should start with PubMed and then search CINAHL. EconLit may also be useful.
• For questions relating to psychology or psychiatry, you should start with PubMed, but then also search PsycINFO.
• If you want to see how many times articles have been cited, you would want to search Web of Science.

It depends on the type of question you’re asking and how comprehensive you need to be in your literature search. Good starting databases for biomedical topics are PubMed, Web of Science, and often CINAHL. To be comprehensive you usually want to search more than one database, but if your question is more about getting a single definition or textbook answer, one database can be enough.

Yes, the strategies taught for PubMed are generally applicable across other literature databases. Other databases don’t use MeSH (Medical Subject Heading) but may use another kind of controlled vocabulary or set of subject headings (sometimes found under “Thesaurus”).

Google Scholar can be a fine place to start searching and, for biomedical questions, often leads back to PubMed results. Google Scholar has always featured relevance ranking of results, whereas this feature was added to PubMed more recently. However, Google Scholar does not have the kind of advanced searching features that PubMed and other literature databases do, especially use of MeSH terms. Google Scholar casts a wider net than PubMed, likely lowering the precision of search results. At the same time, Google Scholar cannot access all of the journals and databases that UF subscribes to because they are behind a pay wall, so you may miss relevant results as well.

Google Scholar does include a couple of filters (by date and articles vs. case law) but does not allow you filter by article type or many of the other filters aimed at biomedical research available in PubMed.

It is always a balance between the achieving precision and recall. As recall goes up, precision goes down because you are widening your search. If you're looking for something specific or literature of a specific type/quality, then emphasize precision. If you are doing research or need to be comprehensive (to make sure you haven't missed anything), then emphasize recall.

There is no way to be certain that you have found everything, you just have to convince yourself that you've done a thorough job with your searching. Usually, when your searches keep finding the same articles, you've come to a kind of convergence.

• Find MeSH terms representing each of the concepts in your search to ensure that you are finding articles addressing these concepts, even if the authors use different terminology than yours.
• In addition to searching the MeSH database, you can find useful MeSH terms by looking at which MeSH terms have been applied to articles you already know are relevant (Expand the drop-down under the article’s abstract to see a list of MeSH terms).
• Supplement MeSH terms with keywords and phrases when there’s not a MeSH term that matches closely with your concept.
• MeSH is not effective for searching the very newest literature in PubMed, before it has been fully indexed (sometimes a delay of several months).

MeSH terms are assigned to articles by librarians at the National Library of Medicine. Librarians will assign an article 8-18 subject headings and 2-5 of these will be major topic headings. The articles entered into PubMed most recently will not have MeSH headings yet.

If you are using MeSH terms, it may not be necessary to include words that address the same concepts as your MeSH terms. However, if you’re searching the newest literature (that doesn’t yet have MeSH terms) or if there is no term that fits with your concept, it is important to include as many synonym keywords as possible.

Searching is often an iterative process; you may find terms from relevant results in an initial search and use these to create a new search. If you’ve already explored which terminology to use and have connected it correctly with ANDs, ORs, and NOTs it is possible to have one or two big searches that are comprehensive in your database.

Filters and limits can be a useful way to narrow results. Some of the most commonly used filters are:
• Date
• Language
• Article type (e.g. limit to only systematic reviews)
As with other elements of searching, whether to use filters and limits depends on the purpose of your search. For example, using a date limit may cause you to lose older literature that could be very relevant; however, if you want the most current knowledge, date limits are very useful.

Knowing if you’ve gotten most of what’s out there on your topic is a matter of trial and error. You can get a sense of what a reasonable number of results might be from a particular database by knowing about the scope and contents of the database.

For instance, if I did a search on treatment of sinusitis in PubMed and only got 50 results, I would be worried about my recall because I know PubMed has over 20 million citations and sinusitis treatment is not a particularly rare or specialized topic.

Some of this depends on how well you know your topic. If you know that it’s not a highly specialized topic but you aren’t getting a lot of results, you may need to revise your search strategy. On the other hand, if you have a topic that really hasn’t been researched much, you many need to broaden your search to other databases or resources. For instance, clinicaltrials.gov allows you to search information about ongoing clinical trials, which may not have any published literature yet. You may also want to reframe your question to address a broader topic or get at something slightly different.

Relevance of results simply refers to how well they answer the searcher’s question. Because an article is relevant to the question asked does not necessarily mean that is a high-quality article. One measure of article quality (or at least importance in the field) is how many times it has been cited by others. The journal in which the article is published, and its impact factor is another related indicator of article quality. Another indication of article credibility is where it fits within the hierarchy of evidence (with systematic reviews and meta-analyses at the top and case studies and opinions at the bottom).

You can export citations from PubMed (using “Send To” > File; Format: MEDLINE) and import them into a citation management program like EndNote or RefWorks. These programs allow you to create bibliographies and generate in-text citations.

If our library doesn’t have access to an article you can request it by interlibrary loan, and we will get it for you, for free, from another library. From within PubMed make a request directly (using “Send To” > Order) by signing up for the free Loansome Doc service. You can also make the request directly through the HSC Library from the Quick Link “ILLiad.”

A vital element to your future practice as physicians is lifelong learning, which includes your ability to continue studying new medical advances that occur after your medical training. Being able to search and evaluate the literature, the record of those new medical advances and new understandings or old practices, means you will be able to engage with your profession, even as it changes. Both evidence-based medicine and medical research require that you be able to search and evaluate the literature for specific clinical and research questions.