library@imperial.ac.uk
+44 (0)20 7594 8810
Limits and filters do what their names suggest. They can be added to a set of search results to further refine them by various criteria such as publication date, age group or study type, discarding records that do not match the criteria.
It is tempting to user limits and filters to reduce the number of results and save time screening, but as you will see in this section they risk removing relevant results without you realising or introduce bias into your review, weakening its quality.
In many cases, an effective search built from strongly defined PICO or other question framework will produce a manageable set of results that requires no limiting or filtering. Preferably, they should be used only when you have an otherwise unmanageable set of results, which is unfortunately unavoidable with some research questions.
Limits offer the option to refine your search according to criteria that are difficult to search for with keywords and subject headings, e.g. the type of publication, publication date or language of the article.
Databases offer many options for limiting results and while these may help you narrow or focus your results when doing a quick or casual search, most of them they are not reliable enough for systematic literature searching. As they generally rely on subject headings only, using them will likely cause you to miss new articles that don't yet have subject headings applied to their records.
To use a limit, you apply it to the end of your search, to the final search line. If you need to use more than one limit, apply limit 1 to the final search line. On this resulting new line, then apply limit 2. This allows the search history to clearly show the effect of each separate limit on the number of results.
Filters are search term combinations designed by librarians and information specialists. They involve a complex, standalone search strategy of keywords, subject headings and other indexing terms and can be added to your strategy to help further limit your final search results to specific methodologies, study types, populations or subject areas. They are far more comphrensive and reliable than the similar limits suggested in databases.
However, like database limits, filters should be used with caution as they also risk removal of relevant results. While using a filter to identify qualitative research seems helpful, study types and methodologies are often not well or variably described by authors and indexing via subject headings is not always consistent. This can make a filter miss identifcation and retrieval of relevant results. However, in many cases a filter is likely to be more sophisticated and effective that what a non-professional searcher can design for concepts such as geographic areas, age groups or subjects like artificial intelligence, quality of life, cost effectiveness, etc.
If possible use a "validated" filter. The most successful and well validated are for randomised controlled trials. Validated filters, which have been tested against gold-standard sets of bibliographic records, are also available for other types of study designs and topics.
Specialised versions of filters have been designed for each database. For instance, the filter to find RCTs in PubMed is slightly different from the filter for RCTs in MEDLINE via Ovid and will not work properly in Ovid, so you must use a filter that has been designed for each database you are using. You will find that most filters are have versions available for multiple databases/platforms.
If you are unsure what filter(s) are suitable for your strategy or wish to adapt or translate a filter, please seek advice from a Medical Subject Librarian.
If you use a filter in your search strategy - either by applying it directly or adapting it, you must ensure it is cited and included in the reference list.
You do not have to completely understand exactly where filters are searching or how they work, but you do need to make sure you are copying and pasting them correctly and putting them into the right place in your search strategy. Always check you are using the filter designed for use on the database you are searching.
The example below shows the Cochrane Highly Sensitive Search Stategy for indentifying randomized trials in MEDLINE; sensitivity maximising version (2023 revision); Ovid format (p. 69-70).
Line 20 is the final line of systematic search strategy. The filter starts on line 21. Lines 21- 28 add the additional concept of RCTs to the strategy and are attempting to find all potential RCTs in the database, about any subject. They are then grouped together in line 29. Line 30 then attempts to find everything in the database that is about animals that doesn't also involve humans. Line 31 then removes everything involving animals found in line 30 from line 29 (all potential RCTs found in the database) leaving only human studies, studies with both humans and animals and studies where neither humans or animals have been specified as test subjects. Line 32 takes the final line of the search strategy (20) and combines it with line 31 - leaving only the results that appear to be RCTs and are not about an animal population.
20. 6 AND 11 AND 19
21. exp randomized controlled trial/
22. controlled clinical trial.pt.
23. randomized.ab.
24. placebo.ab.
25. drug therapy.fs
26. randomly.ab.
27. trial.ab.
28. groups.ab.
29. 21 OR 22 OR 23 OR 24 OR 25 OR 26 OR 27 OR 28
30. exp animals/ NOT humans.sh.
31. 29 NOT 30
32. 20 AND 31
library@imperial.ac.uk
+44 (0)20 7594 8810