4. How do I search for the evidence?

4.7. Refining your search

5.7 Refining your search

Searching databases is an iterative process.

As you review your results, you may decide to refine your search strategy to include, exclude or amend some search terms, limits and/or filters. This is a normal and necessary part of the literature searching process.

Here's some advice to help you address the most common search problems.

Firstly, if you already know of key papers in your field, check that they have been found by your search. If they have not, consider revising your search strategy until they are found.

If you have too many hits: narrow your search

If a carefully conducted search yields lots of results, this suggests your area of interest is complex and researched widely.

Use search terms that are as precise as possible.

Use the powerful ‘AND’ operator to refine your search by adding extra search topics.

Reduce the parameters of your search by selecting publication year, language, publication type, etc. There is a full range of limits, but use some with caution, for example, restricting your search to articles in English is an arbitrary measure, probably excluding excellent research.

Some databases have a 'major heading' option for subject heading terms, restricting the search to articles with your term as a main subject. Again, use this with caution as you may miss some excellent articles.

Some databases have subheadings, e.g. drug therapy, surgery and aetiology in MEDLINE. These allow you to qualify subject headings to limit them to specified facets of research. Be cautious with subheadings – they are not always consistently applied, and you may miss relevant articles.

Apply ‘ready-made’ search filters to find the right data. Some databases have them as Limits, or you can use methodological filters for systematic reviews, meta-analyses, etc. Validated search filters, e.g. for different study types such as RCTs and systematic reviews, are available on the InterTASC website .

.

If you have too many irrelevant hits

You will retrieve a higher proportion of relevant articles if you search with subject headings rather than with free text alternatives. But this does risk missing new articles that have not yet been given subject headings, e.g. in MEDLINE and PubMed.

Use the subject tree/index to find more precise subject headings.

See above.

It can mean you miss relevant results.


If you would like more hits: broaden your search

It may seem obvious, but incorrect spelling, particularly in free text searching, will reduce the number of results. It’s easy to spell cattle with three 't's, for example! Beware of alternative spellings during textword/keyword searches (e.g. behaviour/behavior; immunisation/immunization).

Subject headings are hierarchical; a broad term has narrower terms under it. This may seem an odd term, but choosing a search heading and searching it and all the narrower terms is called exploding the subject heading. Subject headings are often described with the visualisation of a tree with branches. The broad terms are the large branches and the narrower terms are smaller branches from it. An example of this is PubMed’s MeSH subject headings, the MeSH Tree. 

Select ‘All subheadings’ when you are presented with the option, because the subheading system is not entirely reliable.

Most database content is international, so if your search terms do not map to any appropriate subject headings, think of North American or other equivalents.

Look at the subject headings tagged onto a relevant article. Use those terms to expand the scope of your search.

Make sure you are using free text (or keyword) searches as well as subject heading searches, as some concepts will not be captured by subject headings, but will appear in the title and abstracts of publications.

Use the truncation symbol (the symbol can vary between databases, but the most common one is *) in your textword/keyword search to retrieve words with a common root. ‘Tubercul*’ will bring up tuberculosis, tuberculin, tubercule, etc.

No database is complete. In addition to CAB Abstracts and MEDLINE or PubMed, try the Biosis Citation Index, Web of Science, etc.

Especially ones that don’t influence the quality or relevance of search results (for example, ‘abstract only’).