Systematic Reviews: Search Strategy

1. Begin by defining your research question. You should identify who and what are the population, interventions, comparisons, outcomes (PICO) as well as the study design and study characteristics you are interested in.

2. Write down your keyword concepts from your PICO question. Identify the MEDLINE MeSH headings used for your keywords. Your initial search strategy may change as you grow as find new or related additional keywords and concepts.

The PICO model can help you formulate a good clinical question. Sometimes it's referred to as PICO-T, containing an optional 5th factor. 

Although PICO is the most widely used framework for generating a question, there are some situations in which you might find that a different framework is more appropriate. Here are a few alternatives.

• PEO: Population, Exposure, Outcome (Useful for public health and qualitative studies)
• SPICE: Setting, Perspective, Intervention, Comparison, Outcome (Useful for qualitative studies and evaluating projects)
• SPIDER: Sample, Phenomenon of Interest, Design, Evaluation, Research Type (Useful for topics focused on samples rather than populations)
• PIPOH: Population, Intervention, Profession, Outcome, Health Care Setting (Useful for screening and creating practice guidelines)
• PICo: Population, Intervention, Context
• PIT or PICTO: Population, Index test, (Comparator,) Target condition, (Outcome) (Used for diagnostic test accuracy)
• ECLIPSe: Expectation, Client group, Location, Impact, Professionals, Service (Useful for studying outcomes of policies or services)

PRISMA Items 6-7 are:

• "Specify all databases, registers, websites, organizations, reference lists and other source searched or consulted to identify studies. Specify the date when each source was last searched or consulted."
• "Present the full search strategies for all databases, registers and websites, including any filters and limits used."

The PRISMA-Search checklist provides further guidance on how to report the search strategies you use. 

• Use Boolean operators (AND, OR, NOT) to connect keywords.
• Use truncation: If you place an asterisk * (called a wildcard) at the end of a series of letters, the database will search for all words that begin with that series of letters. So the keyword adolescen* will search for adolescence, adolescent, and adolescents. This is useful for capturing the singular and plural versions of words.
• You can add field codes to narrow down your results. For example, in PubMed, searching nursing[tiab] tells the database to search for the keyword "nursing" but only in the titles and abstracts of articles, rather than the full text. 

Sometimes, you don't need to reinvent the wheel. You might find a previous systematic review on your topic that shares the search strategy used. 

• PubMed lists >4,500 reviews that have followed the PRISMA format and in general these will include an example of at least one database search.
• Another source for search strategies is the PROSPERO database of review protocols, type keywords into the search box to see if there are any reviews similar to your topic.
• Search the Dissertations and Theses Online database for completed Doctoral research. The peer-review process for a dissertation is different than for a published journal article, and may not have been subjected to independent scrutiny, however a good PhD dissertation or thesis should always include a literature review and this can be a good source for ideas of resources to search.
• Search MEDLINE or PubMed and use the "Publication type" limit for systematic review to limit the results to just this type of review article. Or search PubMed Clinical Queries using simple keywords and look in the center column of results for a list of recent systematic reviews.

You might also look for a validated "search hedge":

• The websites of the McMaster University Health Information Research Unit, and the University of York Centre for Reviews and Dissemination, may list a pre-existing evidence-based database search filter that can identify studies on your topic.

What types of studies will you be looking for? Certain study designs are better for answering particular question types. The Appropriate Study Designs column lists the study designs best suited to each question type, in order of utility.