Databases for systematic literature searches
MEDLINE AND/OR EMBASE?
Systematic literature searches are an essential tool of evidence based medicine and as such unavoidable in almost every part of medical research.
Their aim is to identify as much available evidence regarding a specific scientific question as possible, by use of a focused, pre-defined and unbiased search strategy.
Two of the most commonly used databases are MEDLINE and EMBASE where MEDLINE is usually preferred due to free access.
Key facts of MEDLINE and EMBASE
As the aim of a systematic literature search is to find as much published evidence as possible, it is advisable to first consider if searches limited to one database, i.e. only in MEDLINE OR EMBASE, are sufficient.
However, many scientists are often unaware of the structural and content-related differences between both databases. Therefore, let’s first have a look at the key facts of MEDLINE and EMBASE:
|Over 23 million references to journal articles||Over 31 million indexed records|
|More than 5600 journals||More than 8500 indexed peer-reviewed journals|
|1946 to present with some older material||Biomedical literature from 1947 to present|
|Indexed with NLM Medical Subject Headings (MeSH)||Elsevier Life Science thesaurus Emtree|
|MeSH is updated once a year||Emtree is updated every three months|
|Free access via PubMed||Access-fee required|
The first fun fact is: All journals listed in MEDLINE are also registered in EMBASE, which means that 2900 journals are unique to EMBASE.
You might ask yourself “Why search in MEDLINE at all, if EMBASE covers it all?”.
“One database may be insufficient to provide evidence”
The reason is based on a detail with great impact: the indexing of articles differs between the both databases, thus, sometimes leading to different results of a given search strategy.
In 2005, Wilkins and colleagues ran an experiment: they performed a search for common family medicine diagnoses in both databases and came up with only 5% of overlap in the results)1.
The experiment impressively proves that solely relying on the search result of one database may be insufficient to provide evidence based answers to scientific questions.
But what causes such differences between both databases?
For one, the indexing of publications within each thesaurus itself.
This task is done by humans, i.e. experts in their fields who index each publication for MEDLINE (in MeSH) or EMBASE (in Emtree). As those experts are not the same persons for each database and humans naturally differ in their opinions, perceptions and their approaches of indexing results, the basis of the structures of EMBASE and MEDLINE can consequently vary as well.
MeSH (used in MEDLINE) is the controlled vocabulary thesaurus of the National Library of Medicine and consists of sets of terms naming descriptors in a hierarchical structure. These descriptors are arranged in both, an alphabetic and a hierarchical structure2.
Emtree (used in EMBASE) is quite similar and was modeled based on MeSH in 1988. It includes a range of terms for drugs, diseases, medical devices and essential life science concepts.
A good overview of the differences of MeSH and Emtree can be found in Elsevier’s whitepaper “A Comparison of Emtree® and MeSH®”3.
In addition, there is a temporal component to the indexing differences: Emtree is used by EMBASE for all articles dating back to 1988, while MeSH-terms in MEDLINE are indexed for even older publications.
“Choose wisely – depending on your field of research”
It is important to note that each database has some degree of specialization.
For instance, while EMBASE has listed more recent medicines and novel drugs as terms in Emtree, MeSH offers a broader range of subheadings.
If you conduct a literature search in EMBASE or MEDLINE, you should be aware of the differences in the establishment of subject headings.
MEDLINE contains more than 26.000 MeSH-terms with 83 subheadings and 213.000 synonyms whereas EMBASE has over 60.000 Emtree-terms (of which more than 30.000 are drug and chemicals) with 78 subheadings (64 are drug subheadings including 47 routes of drug administration) and 260.000 synonyms (over 175.000 drug synonyms).
You see that EMBASE focuses more on drugs and chemicals whereas MEDLINE offers a lot of publications in the fields of dentistry, nursing and veterinary medicine.
So choose wisely – depending on your field of research.
For instance – what if you are not looking for a clinical study with a drug, but rather with a non-drug intervention?
Michaleff et al. rated PubMed on the third and EMBASE on the fourth rank of databases when it comes to searching for randomized controlled trials in the field of physiotherapy.
They ranked the Cochrane Central Register of Controlled Trials (CENTRAL) first and a physiotherapy-specific database (Pedro) second4.
In this search scenario, the evidence based medicine quest for identifying all available evidence suddenly got more complicated.
Research institutions but also regulatory and reimbursement agencies, thus, tend to request a search at least in MEDLINE AND EMBASE (and quite often now CENTRAL as well).
“Two databases are still better than one”
In summary, every database has its own advantages and disadvantages inherent to its structure. The literature gateway Ovid therefore offers the following reasons why a MEDLINE-search is still important5:
Picture: © Dimitrios / Fotolia.com
- Wilkins T, Gillies RA and Davies K. EMBASE versus MEDLINE for family medicine searches: can MEDLINE searches find the forest or a tree? Canadian Family Physician. 2005;51(6):848-9.
- U.S. National Library of Medicine. Fact Sheet Medical Subject Headings (MeSH). https://www.nlm.nih.gov/pubs/factsheets/mesh.html.
- Elsevier R&D Solutions. A Comparison of Emtree and MeSH. 2015. Go to pdf.
- Michaleff ZA, Costa LO, Moseley AM, Maher CG, Elkins MR, Herbert RD, et al. CENTRAL, PEDro, PubMed, and EMBASE are the most comprehensive databases indexing randomized controlled trials of physical therapy interventions. Physical therapy. 2011;91(2):190-7.
- Joint Information Systems Committee (JISC). Why You Need Embase and Ovid MEDLINE. Go to pdf.