PubMed is a free database accessing the MEDLINE database of citations, abstracts and some full text articles on life sciences and biomedical topics. The United States National Library of Medicine (NLM) at the National Institutes of Health (NIH) maintains PubMed as part of the Entrez information retrieval system. Listing an article or journal in PubMed is not endorsement. In addition to MEDLINE, PubMed also offers access to
Many PubMed citations contain links to full text articles which are freely available, often in PubMed Central. In late 2007, President George W. Bush signed the Consolidated Appropriations Act of 2007 (H.R. 2764) into law; this law included a provision requiring the NIH to modify its policies and require inclusion into PubMed Central complete electronic copies of their peer-reviewed research and findings from its funded research. This is the first time the US government has required an agency to provide open access to research and is an evolution from the 2005 policy, in which the NIH asked researchers to voluntarily add their research to PubMed Central.[1] With an effective date of 7 April 2008, the Department of Health and Human Services gave notice: "The Director of the National Institutes of Health shall require that all investigators funded by the NIH submit or have submitted for them to the National Library of Medicine's PubMed Central an electronic version of their final, peer-reviewed manuscripts upon acceptance for publication, to be made publicly available no later than 12 months after the official date of publication: Provided, That the NIH shall implement the public access policy in a manner consistent with copyright law".[2]
The National Library of Medicine also leases the MEDLINE information to a number of private vendors such as Ovid and SilverPlatter – as well as many other vendors. PubMed has been available free on the Internet since the mid-1990s.
Information about the journals indexed in PubMed is found in its Journals Database, searchable by subject or journal title, Title Abbreviation, the NLM ID (NLM's unique journal identifier), the ISO abbreviation, and both the print and electronic International Standard Serial Numbers (pISSN and eISSN). The database includes all journals in all Entrez databases.
As of 16 May 2010[update], PubMed has approximately 19.8 million citations going back to the year 1865.[3] To see the current size of the database simply type "1800:2100[dp]" into the search bar and click "search".
Contents |
For comprehensive, optimal searching in PubMed, it is necessary to have a thorough understanding of its core component, MEDLINE, and especially of the MeSH (Medical Subject Headings) controlled vocabulary used to index MEDLINE articles.
The new PubMed interface, launched in October 2009, encourages the use of quick, Google-like search formulations:
Quick, simple telegram-style search formulations can also be used, and they generally produce acceptable results.[4] PubMed automatically links textwords to relevant MeSH terms. Aspects of the question can then be added successively, in a Google-like fashion, until a number of 'hits' judged manageable is achieved. No knowledge of actual MeSH terms, Boolean operators, English or American spelling, ‘nesting’, or record-fields is required. PubMed's intelligent search algorithm does (or implies) this in the background. Examples of such simple telegram-style questions and results they produce on PubMed:
Telegram-style question in PubMed search window: radial head fractures randomized
Result: 9 records found, one[5] judged highly relevant
Telegram-style question in PubMed search window: glasziou fractures bmj 2007
Result: 1 record (the target) found[6]
Telegram-style question in PubMed search window: vitreous body time death review
Result: 8 records found, several relevant, e.g. Madea/Rödig (2006)[7]
A special feature of PubMed is the 'Clinical Queries' and 'Systematic Reviews' option which can be used to identify more relevant (robust) studies by automatically applying study-type 'filters’ to a search. This feature was updated and re-designed in June 2010.
After a quick search, references which are judged particularly relevant can be marked and 'related articles' can be identified. If relevant, several studies can be selected and 'related articles' to all of them can be generated. The 'related articles' are then listed on order or 'relatedness' (not chronological order). To create the list of 'related articles' PubMed compares words from the Title and Abstract of each citation, as well as the MeSH headings assigned, using a powerful word-weighted algorithm.[8]
A strong feature of PubMed is its ability to automatically link to Medical Subject Headings (MeSH) and Subheadings. Examples would be: 'bad breath' links to (and includes in the search) 'Halitosis', 'Writers cramp' to 'focal dystonia', 'breast cancer' to 'breast neoplasms'. Where appropriate, these MeSH terms are automatically 'expanded'. Terms like 'nursing' are automatically linked to 'Nursing [Mesh]' or 'Nursing [Subheading]'. This important feature makes PubMed searches automatically more sensitive and avoids false-negative (missed) hits by compensating for the diversity of medical terminology.
(For a complete list of tags, see Search Field Descriptions and Tags)
Search-field tags can be used for searching PubMed, some of the most common being:
It should be noted that using field qualifiers automatically disables PubMed's 'mapping' function.
- Note also with the [dp] tag that this searches for publications by both online publication date and print date. Therefore one paper can be associated with two publication years. For example "Schizophrenia 2007[dp]" will return some of the same papers as "Schizophrenia 2008[dp]".
There are three Boolean operators: AND (intersection); OR (union); NOT (exclusion). NOT should be used with care as it may generate 'false-negative' results. The AND operator is assumed by default.
For example:
will yield a single reference, and is the equivalent of
whereas
will yield hundreds of thousands of articles, including all articles in volume 78 of any journal, all articles on page 5275 of any journal, and all articles published in 1981. Note that if tags are not used, PubMed will often attempt to guess at tags and assume the AND operator. For example:
will be translated by PubMed into
which will be correct. Whereas
will be translated by PubMed into the equivalent of
causing PubMed to return the three publications having PMID 78, 5275, and 1981, which was not the intended search result in this case.
All Boolean operators are processed in a left-to-right sequence. The order in which PubMed processes a search statement can be changed by enclosing concepts in parentheses. The terms inside the parentheses are processed first as a unit and then incorporated into the overall strategy.
For example:
is processed by ORing the search (response element OR promoter) first and then ANDing the resulting set of documents with g1p3.
The new PubMed interface, launched in October 2009, with its simple search window, encourages focused search formulations (instead of complex search strategies) when appropriate. It may disourage use of some of the alternative interfaces listed below which offered this simple search option at their opening search page.