The first step is to define a clear and well-defined research question. To make the question searchable, you need to identify the main concepts of the question, and then create different search blocks. Read more on the Search Tips page.
Use a wide range of search terms for each concept in the different search blocks. You need to search with both free text terms and controlled terms, which some databases have in searchable thesauri.
Structure your research question
Start by considering the various components that make up your research question. There are a few different standardised acronyms that can be helpful. A common example is PICO:
- P - Patients, Population or Problem
- I - Intervention or Exposure
- C - Comparison
- O - Outcome
PICO is often used in medicine, but works well for other subjects. Sometimes not all aspects are used, for example if comparison is missing or too vague, PIO can be more useful.
Not all aspects are suitable for inclusion in the search query, geographical areas can for example become problematic. It may be better to include the aspects in your selection criteria instead. See the Screening section below.
Example
- P: Crops of Zea Mays
- I: Cover Crop Treatment
- C: No CCT (comparison with other year or other location)
- O: Yields
Selection criteria:
- Yield records came from corn following a cover crop treatment, and corn following no cover.
- Yields were reported in more than one year or location.
- Enough information was provided to compute study variances.
- The studies were conducted in the United States or Canada.
Example from: Marcillo, GS & Miguez, FE (2017) Corn yield response to winter cover crops: An updated meta-analysis. Journal of Soil and Water Conservation 72(3): 226-239. https://doi.org/10.2489/jswc.72.3.226.
Tip!
There are other acronyms than PICO, read more at Uppsala University Library's guide Systematic reviews.
Test your search
Select one database to test and refine your search in, to ensure that it captures relevant material before proceeding to do searches in all selected databases. An easy way to test whether the search captures what you want to find is to select a number of key articles that should be include in your search. If they appear in the results, you are well on your way, if not, you need to go back and adjust your search.
A common perception when conducting systematic searches is that the number of hits is large, and that much of the material included is outside the scope of the review. It is important to remember that a systematic search needs to be broad enough to capture all relevant studies. There are studies that show that the number of selected articles after screening is usually about 2-4% of the total number of hits from the search, see, for example, the article "Precision of healthcare systematic review searches in a cross-sectional sample". If you still feel that the search gives too extensive a result, you may rather need to go back and adjust your research question.
Tip!
When conducting a systematic review, do not forget that the search and its results are the main part of the data collection.