Searching is an iterative process; you're unlikely to create the perfect search on the first try. Follow the steps below to create a usable search string from a topic or research question.
Step 1: Extract the most relevant concepts from your topic or research question.
PICO is one framework used to help extract relevant concepts from a clinical research question.
P = Population / Problem / Patient
I = Intervention / Issue
C = Comparison / Control
O = Outcome
PICO is a tool to help you break down the question into concepts; whether or not you use PICO is not all that important. The important thing is to break down your research question into its component parts, one way or another.
Step 2: Term harvesting
Next, brainstorm alternative ways of phrasing each of your concepts. The goal here is for you to be able to use these alternative phrasings to create one high quality search which you can use to find the vast majority of relevant literature. Consider: synonyms, variant spellings, singular and plural forms, broader and more specific concepts, acronyms, and archaic terminology.
Step 3: Weave together your search string
With your lists of alternative phrasings in hand, you can now weave together your search string using boolean operators to connect your concepts. Your search string should look along the lines of:
(concept1 OR synonym) AND (concept2 OR synonym) AND (concept3 OR synonym)
Step 4: Search & iterate
As you search, you may notice additional phrasings for concepts which you hadn't originally considered. Add them into your search, within the appropriate concept group. If you are searching in PubMed, you may find it useful to include Mesh terms (PubMed's controlled vocabulary) in your search string. If you are finding too few results, you may want to remove one of your concept groups. If you are finding too many results, you may want to use the database's filters.