Using filters to limit to specific study types when searching bibliographic databases

What are filters and why would I want to know about them?

Filters are tried and tested search strategies that help you narrow down your search, usually to identify articles of a particular level of evidence (e.g. randomised control trials - rcts, systematic reviews), or types of clinical query (e.g. diagnosis, etiology, therapy) or in a specific population (e.g. paediatrics).

The advantage of using a filter is that you benefit from the expertise of people who do this for a living: filters are often written by searchers working for organisations such as the Cochrane Collaboration or SIGN, or by researchers with many hundreds of hours search experience.

Where can I find filters?

Filters are often written up as journal articles. This is also a good place to look if you are interested in finding out about the research process behind writing an evidence-based filter.

BMJ Best Practice, the resource from the BMJ, publishes the filters it uses to identify its evidence.

SIGN also publish their filters on their website (as part of SIGN 50)

An example of a population filter (in this case for paediatrics) can be found at the BestBets website.

However, the filters we are going to look at in this guide are those written by RB Haynes et al of the McMaster University Health Information Research Unit, which are available as Clinical Queries limits within the OVID Medline and Embase databases.

Sensitive or specific?

If you look at the available filters, you will notice that there are often sets of filters which differ in that one is designated “sensitive”, one “specific” and perhaps another that is “optimized”.

A sensitive filter is designed to pick up as many relevant articles as possible and miss as little as possible (ideally nothing relevant is missed). The downside to this is that your results will probably contain many irrelevant results and you will have to examine each one to decide if it is relevant or not. Researchers working on systematic reviews use very sensitive filters as it is vital to find all relevant articles on their topic.

A specific filter is designed to return very relevant results. However, if it is too specific then you run the risk of missing relevant articles. The advantage is that you will have fewer irrelevant results to weed out when you are viewing your results.

There is a trade off between sensitivity and specificity and the purpose of your search should be borne in mind when deciding which one to use.

An optimized filter is designed to be somewhere between the two – specific enough to return mostly relevant results, but sensitive enough not to miss too much.

Searching using the Clinical Queries limits

The McMaster project filters are available as a set of Clinical Query limits via the Additional Limits button.

As it is a limit, it doesn’t display the search strategy used in the filter so you can’t see what keywords etc are being searched for but you can view the search strategies here. The eventual results will be the same as they would be if you had copied in the filter search strategy line by line and combined it with your search.

Searching using a filter via the Clinical Queries limit

To demonstrate the use of filters we will run a search in Medline for randomised control trials on the use of hypnosis in dentistry.

First we construct a strategy for the topics “hypnosis” and “dentistry”. For demonstration purposes this is a simple three line search. It is normally preferable to search using key words as well as subject headings.

To help to identify the rcts among the 228 articles our search found we can apply a filter.

1. Click on “Additional Limits” (see screenshot above) and scroll through the available Clinical Queries until you get to Therapy (specificity). The therapy filters look for the best study design for answering questions about the effectiveness of therapies, i.e. rcts. Other question types will be best answered by other study types and the filters will reflect this. 
2. Click on the filter you want to apply and then scroll to the top and click on the Limit a Search button.

Results

• Applying the specific filter reduced the number of results to 6, all of which were RCTs. 
• Running the sensitive filter produced 38 results, 6 of which were RCTs; 8 further controlled clinical trials, some of which may have been randomised but it was not clear from the record; 3 systematic reviews; and 21 irrelevant results
• Running the optimised filter produced 12 results – the 6 RCTs found in the previous filters; 2 controlled clinical trials; 3 systematic reviews; and 1 irrelevant result.

Applying filters as a search strategy

If you want to use a filter other than the Clinical Queries options, or if you want to have the lines of the filter visible in your search strategy, you can type or (preferably) copy and paste the filter in line by line.

1. First identify the filter you want to use, for example using one of the links above or in the Useful Links below.
2. Make sure that the filter was both written for the database you're wanting to search (e.g. Medline or Embase) AND the search interface that you're using (e.g. OVID or Ebsco)
3. Create your topic based search strategy as normal
4. Copy and paste the filter in one line at a time, clicking on Search after each line
5. Finally, combine the last line of your topic search with the last line of the filter to get your filtered results

Saving a filter

If there's a filter that you use a lot, you may want to save it as a search strategy so that you don't have to enter it from scratch each time.

We'll look at how to do this in OVID but it will work with the Ebsco search interface as well if you're searching CINAHL.

Setting up an OVID personal account

To save searches in OVID you need to have a personal account. This is basically a bit of space for you on the OVID server where you can save searches and set up alerts. Some filters are saved as “expert searches” in this space so check that the filter you're using isn't already saved here first.

To set up a personal account click on the “My Account” button at the top of the screen.

Once you have set up your personal account you can log in using the username and password you chose.

You can use your personal account to save search histories that you might want to use again and set up automatic alerts.

1. Enter the filter line by line as detailed above (this time you don't want to have done a topic search first - we want the search to only be the lines of the filter).
2. Click on the "Save All" button at the bottom of your search
3. Log in to your Personal Account and give your filter a name. Make sure you include the database that it's for as well, either in the name or comment field
4. Click on Save.

Using a saved filter

To add a filter to your search, click on the check box beside it and then click on the Run button at the top of the page. When you return to the main search page you will see that the filter search has been added. Now all you need to do is combine the last line of your search with the last line of the filter using AND.

Click on the “View Saved” button, or the “Saved Searches/Alerts” to view the content of your personal account.

1. Create your topic based search strategy as normal
2. Click on “View Saved” and log in to your OVID personal account
3. Find the filter that you've saved and tick the box on the left hand side of the title
4. Click on the Run button
5. When you return to the main search page you will see that the filter search has been added. Now all you need to do is combine the last line of your search with the last line of the filter using AND.

Useful links:

What is a search filter?
Good overview with examples from Best Evidence Topics (Manchester Royal Infirmary).

The InterTASC Information Specialists’ Sub-Group Search Filter Resource
A collaborative venture to identify, assess and test search filters designed to retrieve research by study design, from the Centre for Reviews and Dissemination, University of York.

Clever searching for evidence
BMJ editorial from 21^st May 2006 on how using filters can help find the needle in the haystack of medical literature. With links to other relevant articles.

McMaster University Health Information Research Unit (HIRU) Hedges Project
Information about search filters for Medline, Embase and PsycInfo