13 Research methods, statistics, critical review and evidence-based practice13.1 Introductory concepts
13.1.5 Formulating clinical questions
3 min read
Written by Resident Doctor Committee · Published
Formulating an answerable PECO(T) question
The PICO format
PICO (population, intervention, comparator, outcome) turns a clinical uncertainty into a focused, answerable and searchable question. It is framed before the literature search begins.
Element | Meaning |
|---|---|
Population | The group of patients being asked about |
Intervention or exposure | The treatment, test, risk factor or service model |
Comparator | The alternative it is weighed against: another treatment, usual care, placebo or nothing |
Outcome | The benefits and harms the patient cares about |
Population should include only features that could change the answer, such as treatment resistance in a clozapine question.
PECO is the exposure-based variant: patient or population, exposure, comparison, outcome. It fits aetiology, harm and some prognostic questions.
Choosing the comparator
For the risk of a drug side effect, the comparator is people with the same condition who are not exposed to the drug. This isolates the drug's contribution and mirrors the patient's decision to start it or not.
Placebo differs from doing nothing, because it carries the context and expectations of treatment.
A descriptive prognosis question may need no comparator.
Outcome and time
A rating-scale change can be statistically significant yet too small to matter, so judge it against the minimal clinically important difference.
State the follow-up when it changes the interpretation, above all for relapse or death.
Related frameworks
Framework | What it adds or uses | Fits |
|---|---|---|
PICOT or PECOT | Time | Questions affected by follow-up duration |
PICOS | Study design | Systematic review eligibility criteria |
SPIDER | Sample, phenomenon of interest, design, evaluation, research type | Qualitative questions |
SPICE | Setting, perspective, intervention, comparison, evaluation | Questions needing an explicit setting and perspective |
Types of clinical question
Background and foreground questions
Background questions ask for general knowledge about a disorder: its features, causes, frequency or mechanism. They frame no population, intervention or exposure, comparison or outcome.
Foreground questions are specific and patient-oriented, naming a defined population, intervention, comparison and outcome so that the answer informs a clinical decision.
Question types and matching designs
Type | What it asks | Best matched design | Typical effect measure |
|---|---|---|---|
Therapy | Does an intervention help? | Randomised controlled trial (RCT), or systematic review of RCTs | Risk ratio, absolute risk reduction, mean difference |
Harm | What adverse effects follow? | Large cohort, linked routine data, pharmacovigilance registry | Risk ratio, incidence |
Aetiology | Does a factor cause the outcome? | Triangulated observational evidence, Mendelian randomisation where relevant | Risk ratio, odds ratio |
Prognosis | Does a factor predict the outcome? | Inception cohort | Absolute risk, hazard ratio |
Diagnosis | How accurate is a test? | Comparison with a reference standard | Sensitivity, specificity, likelihood ratios |
Economic | How do costs compare with outcomes? | Purpose-built economic evaluation | Cost per quality-adjusted life year |
Qualitative | What do people experience, and why? | Interviews, focus groups, observation | Themes, not a numerical effect size |
Prognostic factors need not be causal, so a predictor does not show that changing it would change the outcome.
Event rates in trials
The control event rate and experimental event rate are the percentages of each trial arm who have an event on a dichotomous (yes or no) outcome.
Relative risk is derived from the two rates, but the rates on their own are not measures of benefit.