How to write a PICO question for a meta-analysis

PICO is a framework for turning a clinical idea into a question a review can answer. It stands for Population, Intervention (or exposure), Comparator and Outcome. A review question that is vague in any of the four will produce a search that is too big, a set of studies that cannot be pooled, or a result no one can use.

This guide shows how to write each element, which variants to use for different types of question, and the mistakes that make a PICO unusable.

Write each element precisely

Population: who exactly? Give the condition, the age group, the setting and any severity or stage. Adults with type 2 diabetes and chronic kidney disease is better than diabetic patients.
Intervention or exposure: what is being done, or what are people exposed to? Name the drug class, dose range, device or procedure, and the duration if it matters.
Comparator: placebo, usual care, another treatment or no exposure. If you accept several comparators, say how you will handle them.
Outcome: what is measured, how and when? State the primary and secondary outcomes, the measurement tool and the time window. Name the outcome patients care about, not only a surrogate.
Study design: add which designs you will include (randomized trials only, or also cohort studies). This is sometimes written as PICOS, with S for study design.

Variants for other kinds of question

Exposure and risk (PECO): population, exposure, comparator, outcome. Used for questions about harms and risk factors.
Prognosis (PFO): population, prognostic factor, outcome, with timing and setting.
Diagnosis (PIRD): population, index test, reference standard, diagnosis of interest. A diagnostic accuracy review asks for sensitivity and specificity, not an effect size.
Prevalence (CoCoPop): condition, context and population.
Qualitative or mixed-methods questions (PICo, SPIDER): the sample or population, the phenomenon of interest, and the context or design.

Let SynthGap run the first pass

Enter an area. SynthGap proposes candidate topics, checks each against existing systematic reviews, and lists the primary studies behind it. Start on the Free plan: no card needed, and every run shows every topic it finds.

Four worked examples

These show structure. They are not claims that the questions are unanswered.

Treatment: In adults with heart failure with preserved ejection fraction (P), do SGLT2 inhibitors (I) compared with placebo (C) reduce hospitalisation for heart failure and cardiovascular death (O), in randomized trials (S)?
Exposure: In pregnant women (P), is exposure to a given antidepressant class in the first trimester (E) compared with no exposure (C) associated with congenital malformations (O) in cohort studies (S)?
Diagnosis: In adults presenting to the emergency department with chest pain (P), what is the accuracy of high-sensitivity troponin at a stated threshold (index test) compared with a final adjudicated diagnosis of myocardial infarction (reference standard)?
Prevalence: What is the prevalence of burnout (condition) among resident physicians (population) in hospital settings (context)?

Mistakes that make a PICO unusable

An outcome such as effectiveness or outcomes, with no measure.
A population so broad that the effect would differ across it, such as patients with cancer.
Several interventions bundled together so that nobody can say which one works.
A comparator that is not stated, which makes inclusion decisions arbitrary.
Changing the PICO after seeing the results. Fix it in the protocol and register it.

Questions

Do I need a PICO for a scoping review?

Scoping reviews use a related framework called PCC: population, concept and context. It is broader because the aim is to map evidence, not to estimate an effect.

How narrow should my PICO be?

Narrow enough that the studies can be pooled meaningfully, broad enough that you find enough studies. If you find too few, consider broadening one element, usually the population or the comparator, and write down the reason.

Where do I register my PICO?

In a protocol registered on PROSPERO, if your review qualifies, before data extraction starts. The registration records the question and the planned methods.

Keep reading

General methodological guidance, not medical advice. SynthGap's suggestions are AI-generated from the published literature, so verify them before registering a protocol.