A research question is only as good as its structure. "I want to study whether yoga helps anxiety" is a topic, not a question. To turn it into something you can actually search, screen, and synthesize, you need a framework that forces specificity. PICO is the most famous, but it is not the only one, and using the wrong framework forces your question into a shape it does not fit. Here is when to use which, with the same broad idea (yoga for anxiety) reformulated in each.

PICO: for quantitative intervention questions

PICO stands for Population, Intervention, Comparator, Outcome. It is the default for any question about whether one treatment works better than another — randomized trials, meta-analyses of trials, and most systematic reviews use it.

  • Population: Adults aged 18 to 65 with generalized anxiety disorder (DSM-5 criteria)
  • Intervention: 12-week structured Hatha yoga program, 2 sessions per week
  • Comparator: Usual care (counseling and pharmacotherapy as indicated, no yoga)
  • Outcome: Generalized Anxiety Disorder 7 (GAD-7) score at 12 weeks

The full question reads: "In adults aged 18 to 65 with generalized anxiety disorder, does a 12-week structured Hatha yoga program (2 sessions per week) compared with usual care reduce GAD-7 scores at 12 weeks?" That is a question PubMed can actually answer.

PICO is the right choice when there is a clear intervention being applied and a control to compare against, and the outcome is measurable. Most therapy questions fit.

PECO: for observational and etiology questions

PECO replaces "Intervention" with Exposure. Use it when no one is administering a treatment — you are studying who got exposed to something and what happened to them. Etiology, risk-factor, prognostic, and environmental-epidemiology questions almost always use PECO.

  • Population: Adults aged 30 to 70
  • Exposure: Daily yoga practice for 5 or more years
  • Comparator: No regular yoga or meditation practice
  • Outcome: Incidence of generalized anxiety disorder over 10 years

The question becomes: "In adults aged 30 to 70, is long-term daily yoga practice (5+ years) associated with a lower 10-year incidence of generalized anxiety disorder compared with no regular yoga practice?" This is now a cohort-study question, not a trial question. Different evidence pyramid, different study designs to look for (cohort, case-control, cross-sectional), different risk-of-bias tool (ROBINS-E or Newcastle-Ottawa, not RoB 2).

SPIDER: for qualitative and mixed-methods questions

SPIDER is built for qualitative evidence. It stands for Sample, Phenomenon of Interest, Design, Evaluation, Research type. Note that "Sample" is deliberately not "Population" because qualitative samples are small and purposive, and the word matters when you search.

  • Sample: Adults with anxiety disorders who have practiced yoga for at least 6 months
  • Phenomenon of Interest: Lived experience of yoga as a coping strategy
  • Design: Semi-structured interviews, focus groups, ethnography
  • Evaluation: Themes related to perceived benefit, barriers, adherence, meaning
  • Research type: Qualitative or mixed-methods

The question becomes: "What are the lived experiences of adults with anxiety disorders who use yoga as a coping strategy, and what themes emerge regarding perceived benefit, adherence, and meaning?" This is a fundamentally different question. It cannot be answered with a forest plot, and it should not be forced into PICO just because PICO is familiar.

How to pick: a one-question decision tree

Ask yourself: "Is someone in the study doing something to the participants, and am I comparing it to an alternative?"

  • Yes, with a clear comparator → PICO
  • No, I am observing what naturally happened to them → PECO
  • I want to understand how they experience or make sense of something → SPIDER

If your question wants to know both whether something works (quantitative) and how people experience it (qualitative), you have a mixed-methods question. You can use PICO and SPIDER side by side, or use a hybrid framework like PerSPEcTiF, which is purpose-built for mixed-methods reviews of complex interventions.

Other variants worth knowing

  • PICOT — PICO plus Time frame. Useful when the outcome timing is critical (e.g., 30-day mortality versus 1-year mortality).
  • PICOS — PICO plus Study design. Common in systematic-review protocols when you want to restrict to RCTs only.
  • PIRD — Population, Index test, Reference test, Diagnosis. For diagnostic accuracy studies. PICO does not fit them well.
  • CIMO — Context, Intervention, Mechanism, Outcome. Used in implementation science and realist reviews.

Why this matters for your search

Each element of your framework becomes a concept block in your search string. If you used PICO, you have 4 concept blocks. If you used SPIDER, you have 5. The same question framed as PICO and SPIDER produces very different searches and very different included papers. Choosing the wrong framework can produce a polished-looking review that systematically misses the right evidence.

If you are using ResearchChecker to scope an idea, the tool prompts you to write your question as freely as you want, but you will get better results if you have already structured it in your head using one of these frameworks first.