HOW TO DEVELOP THE SEARCH STRATEGY FOR A SYSTEMATIC REVIEW

Hello all, welcome back after a long break. Hope you have registered your review in PROSPERO by now. We are now moving to the crucial yet less discussed step of a systematic review, which is developing the search terms. Search terms are the engine of a systematic review. If it is in the right direction, it reaches the destination but if it’s in a wrong direction, it can mislead you, and you might not reach your end-point. The search in a systematic review should be as extensive as possible to ensure that all necessary and relevant articles are included. After all, you are generating the evidence!

Before you run the search and retrieve articles, be careful about what and how many search terms you have used. Ask for help if you are unsure about it. Many of my friends have asked me about it, and I have told the same thing which I am going to tell you now…Using more search terms is not wrong but using fewer terms might lead to missing of some pertinent articles. Don’t be scared to use as many terms as possible in the beginning, and later you can delete the irrelevant terms. Let’s see the stepwise procedure.

• Decide your keywords:

You begin by deciding your keywords. Keywords are the words which summarize your topic. You have to use your PICO terms for deciding your keywords. E.g., if my topic is ‘Effect of TENS on spasticity in adults with stroke’ then my keywords will be ‘TENS’ AND ‘Spasticity’ AND ‘Stroke.’ We also call keywords as concepts.

• Expand each keyword:

Now that you have decided your keywords start growing the terms for each concept/keyword. You can do that by looking for synonyms for each word. Add singular as well as plural words, e.g., activity and activities. You can use variant spellings, e.g., Disc and disk. You may also add other related terms to widen your search.  Truncation and wildcards are other smart ways to retrieve related articles. E.g., for truncation is ‘random*,’ if you use this truncation, then it will search all terms related to random, i.e., random or randomised or randomized or randomly, etc. E.g., for wildcards is ‘wom?n’  so it will include articles with both woman or women. Another simple way to look for synonyms is jotting down all the MeSH terms of the keyword from PubMed and Cochrane. I have attached the pictures on how to search for MeSH terms. You can also add synonym by going to the appendix of any Cochrane review done on a similar topic and writing down the words used in its search strategy.

• Tabulate the search terms:

For clarity and convenience while running a search, tabulate all your keywords as shown below using the same example of keywords as given earlier.

Stroke	TENS	Spasticity
cerebrovascular accidents	transcutaneous electric nerve stimulation	stretch reflex
cerebrovascular disease	transcutaneous nerve stimulation	muscle hypertonia
brain ischemia	transcutaneous electrical neurostimulation	spasm
carotid artery disease	percutaneous neural stimulation	clonus
intracranial embolism	micro amperage electrical stimulation	spastic paresis
intracranial thrombosis	transcutaneous electric stimulation	hypermyotone
intracranial hemorrhage		high tone
brain infarction		hypertonia
vertebral artery dissection		hypertonicity
hemiplegia		hyper reflexia
hemiparesis		hyperreflexia
monoparesis		hyper-reflexia
cerebral vascular disorder		spastic reflex
cerebellar vascular disorder		spastic
brainstem vascular disorders
vertebrobasilar vascular disorders
subarachnoid hemorrhage
intracranial aneurysm
infratentorial hemorrhage,
supratentorial hemorrhage
lacunar stroke
intracranial arteriovenous malformations
brain attack
Final result = R1	R2	R3

• Build your search strategy:

Now build your search strategy by joining the terms under each column with Boolean ‘OR’. Suppose after adding all terms, you get the final result as R1, R2, and R3 for each column respectively, then add all of these with Boolean ‘AND,’ i.e., R1 AND R2 AND R3. You can even add ‘NOT’ if you want to exclude a few terms specifically. See the example in the picture below.

• Running the search:

Once you have completed the herculean task of developing the search terms, you can go ahead with running the search in different databases. You should also include conference abstract and other grey literature. Another important source which may have potential articles are trials registers of ongoing and unpublished trials.

• Few tips:

Keep a record of all the databases searched and number of articles retrieved from each database. Save the search strategy and include it in the appendix of your manuscript. Additionally, try to keep a fine balance between precision and comprehensiveness. Which means that you should try to keep your search extensive but not so extensive that you retrieve more irrelevant articles. Don’t worry; it will come with practice and experience. Let me know if you have any muddles while building your search strategy.