Systematic review searching
A systematic review requires:
Time: A systematic review is a rigorous, time-consuming undertaking
A team: You need to work with subject experts, at least one other reviewer for screening citations for relevance, and a librarian who can advise on suitable resources and develop the search strategies, if required. If you are including a meta-analysis, you may also need the help of a statistician. Someone is also required to manage the data and write up the report.
An initial scoping search: This will reveal if a systematic review has already been undertaken on your topic as well as highlight gaps in the knowledge base. A scoping search will also allow you to develop and pilot inclusion/exclusion criteria, and should give you an initial sense of the quantity of existing primary research on your topic (and therefore the scale of the project).
A clearly focused question: This should flow from your initial search. The question should be worthy of an answer and should strike a suitable balance between being too broad or too narrow in scope. If appropriate, use the PICO framework to identify the important concepts in your question.
Clear inclusion/exclusion criteria: These should be based on your question. Consider all aspects of the topic such as age groups, geographic regions, types of study designs, languages, stage of an illness, and the outcome measures that need to be described. Clear eligibility criteria will make it easier to identify relevant articles at the screening stage and prevent you being distracted by interesting but irrelevant studies. Be wary of date range limits in a systematic review. They must be justifiable, not arbitrary.
A comprehensive literature search: This must include a search for unpublished studies as well as published ones. It should target an exhaustive set of relevant databases and include a Google search for grey literature. You might even consider searching the contents pages of all issues of a select set of highly relevant journals. This is called 'hand searching'. Search strategies need to be well documented, including enough detail that they can be replicated. The numbers of results from each database search should also be included.
Critical appraisal: Each study meeting the inclusion criteria must be appraised for quality (internal validity) as well as applicability (external validity or generalisability). A critical appraisal tool such as the CASP RCT checklist can guide you in this.
Good citation management: EndNote is good for taking your search results offline and eventually deduplicating the results from multiple databases. Create a separate EndNote Library for each database searched plus one for citations found via other means (e.g. consulting reference lists or hand searching). Copy all citations into one total library and use the EndNote 'Find Duplicates' command to deduplicate. You will still need to visually remove duplicates, as this process is not perfect. This then becomes your working library.
The systematic review protocol
Systematic reviews should work to a pre-defined plan (or 'protocol') in order to reduce the risk of bias in the process. The protocol guides your process and prevents you from answering a different question to the one you set out to answer.
Writing a protocol
A protocol should describe:
- the question being asked
- the resources and approaches to identifying studies that the authors intend to use
- the inclusion/exclusion criteria against which the studies will be assessed
- the assessment criteria/tools to be applied to gauge individual study quality
- how the data from each study will be extracted and synthesised.
The PRISMA-P checklist is a useful guide to follow when writing a protocol.
Registering your protocol
Registering your protocol in a publicly accessible way will avoid other people duplicating your review. Similarly, it is always a good idea to check these sources ahead of starting out just in case someone else has lodged a review protocol on the same topic.
The main sources of registered protocols are listed here:
Quantitative Appraisal Tools
Qualitative Appraisal Tools
Mixed Methods Appraisal Tools