A research question for a systematic review is a single, precisely scoped question that the whole review is designed to answer, narrow enough to be searchable yet broad enough to matter. A strong question names the exact population, the intervention or exposure, and the outcome, so that it can be turned directly into a search and a set of eligibility criteria.
Why the question carries the whole review
Every later decision inherits the question. The search terms come from it, the eligibility criteria operationalise it, and the synthesis answers it. A question that is too broad returns tens of thousands of records and an unfinishable review; one that is too narrow finds three studies and says nothing useful. Getting this balance right is the most important hour you will spend, and it is why we treat it before anything else in the protocol. A question that cannot be searched cannot be reviewed.
From a vague aim to an answerable question
Start with the framework
Frameworks exist precisely to convert a fuzzy aim into structured elements. For intervention questions, the PICO framework splits the aim into Population, Intervention, Comparator, and Outcome. For other kinds of question, you reach for a different structure entirely, which we set out in question frameworks for systematic reviews. The framework is the tool that does the sharpening.
Check it has not already been answered
Before committing, search for existing reviews on the same question. If a recent, well-conducted review already exists, you either need a different angle or an update. A quick scoping search of the field, and a look at registered protocols, tells you whether your question is genuinely open.
Test it against the FINER qualities
A useful gut check is whether the question is feasible, interesting, novel, ethical, and relevant. In practice that means: can you realistically find and screen the records, will anyone care about the answer, has it not already been settled, and does it address a real gap. A question that passes these is worth registering; one that fails usually needs re-scoping before you write a single search line.
Worked example: from aim to answerable question
Watch a typical aim tighten in three passes. A researcher begins with “Does mindfulness help mental health?” That is a topic, not a question: the population is everyone, the intervention is undefined, and the outcome could be a dozen different things. The first pass names the people and the outcome: “Does mindfulness reduce anxiety in adults?” The second pass commits to a specific intervention format and a comparator: “Do eight-week mindfulness-based stress reduction programmes, compared with a waitlist, reduce anxiety in adults?”
The third pass fixes the measurable outcome and the eligible designs: “In adults with a diagnosed anxiety disorder, do eight-week mindfulness-based stress reduction programmes, compared with waitlist or usual care, reduce anxiety symptom scores at programme end in randomised controlled trials?” That final version names every element a PICO structure needs, which means it can be turned line by line into search concept blocks and eligibility rules. Notice the ambition never shrank; only the vagueness did.
Signs a question is too broad or too narrow
Two failure modes sit at opposite ends, and a pilot search exposes both before they cost you months:
- Too broad. The pilot returns tens of thousands of records, several PICO elements are unstated, and the eligible designs are left open. Such a question promises a review that never finishes screening and a synthesis so heterogeneous that no honest pooled estimate is possible.
- Too narrow. The pilot returns a handful of records, the population is restricted to a single rare subgroup, and the answer, even if found, generalises to almost no one. A review of three small studies rarely justifies the effort and cannot support a firm conclusion.
The fix is asymmetric. A broad question is tightened by adding constraints one at a time; a narrow one is widened by relaxing the most defensible restriction first, usually the date limit or a needlessly specific intervention format, never the core population the review is meant to serve.
Scoping the question to a finishable size
The single most common failure is a question that is too broad. If a pilot search returns an unmanageable volume, tighten one element at a time: narrow the population, restrict the comparator, or limit to a stronger study design. The breadth of the question is also the main driver of how long the work takes, as we explain in how long a systematic review takes. A well-scoped question is not a smaller ambition; it is the difference between a review you finish and one that finishes you.
Once the question reads cleanly, pressure-test it against neighbours. Is it genuinely distinct from an effectiveness review you could pool, or are you really describing a breadth-mapping exercise that wants a different design? Has a recent review already settled it, in which case an update or a fresh angle is the honest move? Reviewers who answer those two questions before drafting the protocol almost never have to re-scope after the search, and they can hand a search specialist a target that is precise enough to build a strategy around. If you would rather have that target stress-tested for you, our protocol service sharpens the question and locks it into the registered methods.