Friday, November 06, 2015

Translating search strategies

Not sure if I borrowed this term, or invented it, but I mean modifying a search strategy for use in another database.   For example, if you have a Medline strategy, what do you need to change to make it work in, say, PsycINFO or Embase?  Other databases might have different features or functionality.

There are two areas where translation is necessary, once you have finalised your strategy.  I have assumed in what follows that you have a Medline strategy, and want to translate it for use in other databases.

One area is search syntax (proximity operators, truncation, and so on), and the other is subject headings.

For example:

Proximity operators.   In Ovid Medline, adj finds your two terms adjacent to each other, so:

heart adj disease 

finds heart disease.

More usefully, perhaps, adjn finds your two terms within n words of each other, in either order.  So:

heart adj2 disease

finds heart disease, disease of the heart.

heart adj2 disease* 

finds those, and also diseases of the heart.

adj and adjn work in Ovid Embase, and also in the HDAS NHS England databases.

In PsycINFO via Ebsco, adjn is Nn.  In Web of Science and the Cochrane Library (via the Wiley interface) it is NEAR/n.

Truncation and wildcards.  In many databases truncation is done with *.  Ovid Medline, Embase and HDAS use * but will also accept $. So, disease* finds disease, diseases, diseas* finds those and diseased.  In many databases, you can use ? and # in the middle of words to stand for one character, or for one character or a space, respectively.  So:

wom?n finds woman, women
p#ediatrics finds paediatrics, pediatrics

You would need to check in your database that * is the truncation symbol, and if you can use wildcards like ?, #.

Phrase searching.   Ovid assumes that two adjacent words are a phrase, but NHS databases, Cochrane, Web of Science and PsycINFO via Ebsco do not and you need to mark phrases "like this", in speech marks/double quotes.

Subject headings.  Your Medline search will (or should!) include MeSH terms.  But your next database might not have MeSH.  Cochrane does use MeSH (but not in all databases), and you need to use the Search Manager to be able to browse headings.  Web of Science has no thesaurus, and PsycINFO and Cinahl have their own.  

In PsycINFO or Cinahl, identify the term that is used for the concept you are searching, by entering the MeSH term or another keyword and exploring the thesaurus.    

Cinahl's thesaurus is, I think, based on MeSH, but the term might differ, or it might be possible to explode it in Medline but not in Cinahl (Diabetes Mellitus, Type 2, is an example).

PsycInfo will have more detailed terms for psychological concepts, but possibly less detailed for other things.  For example, there is only one subject heading for Diabetes, which includes Type 1 and Type 2.   

Embase's thesaurus goes into more detail with regard to things like drug delivery route, and may have different terms for a concept.  For example, type 2 diabetes in Medline is indexed under Diabetes Mellitus, Type 2, and in Embase under non insulin dependent diabetes mellitus.

In a database without a thesaurus, I would ensure I have searched for the MeSH term as free text.  

Finally, I have learned that having a Medline strategy where MeSH and freetext are on separate lines makes translation easier.

No comments: