PICO framework examples for medical researchers
Example: Population: adults with type 2 diabetes; Intervention: GLP-1 receptor agonists; Comparison: standard care or placebo; Outcome: HbA1c change or cardiovascular events.
Build one concept block per PICO element. Synonyms and spellings go inside parentheses with OR; blocks combine with AND. This structure is what librarians mean by a "bulletproof" PubMed strategy.
- Population block: disease terms + MeSH + free-text variants.
- Intervention block: drug class + brand names + trial filters if needed.
- Outcome block: often optional at search stage; apply at screening instead.
Why your systematic review search strategy is too broad
Common causes: single-keyword searches, missing study-type limits where appropriate, no exclusion concepts, and importing every database hit without deduplication.
If PubMed returns tens of thousands of records, your protocol may be fine. However, your search strings need tightening before anyone opens an abstract.
How to refine your PubMed search strategy using a Generative AI co-researcher
Meta-analysis360 generates an initial PubMed string from your planning text, then lets you amend and propagate changes across related database blocks. The LLM acts as an additional reviewer to estimate counts before import so you know whether to narrow further.
Pair search refinement with Generative AI screening when you already exported a large library. Fix forward for updates, and triage what you have now with increased speed and quality.
“A search that returns 40,000 records is not a victory: it is a planning problem. PICO structuring and count checks before export prevent teams from drowning in screening.”
Generate and refine PubMed search strings
Turn your PICO into amendable database strings and estimate result counts before import.
Common questions
Should I limit to RCTs in PubMed?
Only if your protocol restricts study design. Over-filtering at search can miss eligible quasi-experimental or observational studies. Match limits to your PROSPERO eligibility.
How many results is too many?
There is no universal cutoff: it depends on team size and timeline. If dual manual screening would exceed your deadline, refine the search or plan AI-assisted first-pass screening.