Try following this 'PICO' method for identifying the components of your research question. Use background sources and preliminary database searches to help you define the scope and definition of each component, thereby helping you develop an effective research question (one that is sufficiently specific, focused and answerable), while also identifying appropriate vocabulary to use in your searches.
P is for the Patient or the Problem (e.g. a disease or condition) or the Population of interest
I is for Intervention (e.g. therapy, procedure, drug, exposure, test, strategy)
C is for Comparison (e.g. compared to an alternative intervention or control)
O is for Outcome (i.e. the consequence, effect or improvement of interest and the measurement thereof)
Not all research questions will fit into the PICO strategy (e.g. questions asking about etiology, diagnosis, prognosis, risk factors, frequency), so it's helpful also to consider the:
Type of question you're asking, and also the
Type of research study most likely to provide your answer (e.g. would it be a systematic review; a meta-analysis; a clinical guideline; or a qualitative study)
To incorporate these latter considerations, "PICO" is sometimes extended to PICOTT.
Your searches for the best available evidence will be most productive if you a) search in several different databases, b) try a variety of different terms/phrases to express each of the PICO concepts in your research question, and c) take care to evaluate the strength, relevance, currency and quality of the evidence you find.