Coming up with a search strategy

If you're not an information specialist, you may find the sources listed above a bit daunting. Don't worry; it doesn't all have to be done at once. Remember the logical approach of starting to look at the richest source and working down.

Remember that this is something your review group should be able to help you with. Get in touch with them and ask what help they can give you.

Document your search

It's very important to keep an accurate record of what you've searched, when you searched it and how you searched it. It'll help you avoid having to repeat searches and it will help people using your review to appraise how well they think you've minimised bias.

All these details should be documented in the 'Search strategy' section of the text box in RevMan. The only exception is that where you've used the register of your review group, you don't need to write down the strategy your group used. You should, however, explain the search used to retrieve studies from your review group's register.

Keeping it under control

Keeping track of searches can be a challenge. You may find several reports of the same study, and you will probably find the same report of a study in several databases. So you need some way of keeping track of the references you've looked at, and then some way of grouping together all the reports of a single study.

You might like to keep a record of where you found each study, so that you can report how useful different sources were.

Some people use reference management software to do all this, such as ProCite, Reference Manager, EndNote or IdeaList. If you like working with databases this is great, and can save time typing in references later on. Other people prefer printing out citations and writing on them. You're going to need some system for keeping track of which references you think are relevant, which ones you have ordered from the library, which ones you've received the paper for, etc. It's a good idea to keep a note of which studies you have found and rejected. You may well come across them again later and it can be very frustrating to re-read irrelevant records.

Whichever route you take, you'll probably end up with a file with a section for each study. In this section you'll keep the form you use to collect information about the study and all the papers that report that study.

Tips for saving time and effort

Here are a few ways you can save yourself some time and effort. At the risk of being repetitive, the main advice is simply to get some help from an expert. But some other little tips are:

• look at the terms used to index and describe a few studies you already know are relevant to your review, and use these terms in your search strategy
• add new terms to your search strategy and then pilot them on part of the database to see whether you get relevant material, before you run it on the whole database
• use date limits for your search if appropriate. For example, if drugs, surgical techniques or diseases have only been around since a certain date, there's no point searching before then