skills/paper-critical-appraisal/references/appraisal-lenses.md

# Study-design-sensitive appraisal prompts

Apply only the relevant subset. These are analytical prompts, **not** substitute scoring rubrics for RoB 2, ROBINS-I, QUADAS-2, CASP, AMSTAR 2, or GRADE.

## Randomized trial
- Randomization and concealment; baseline balance; blinding where feasible; deviations from interventions; losses/missing outcome data; prespecified versus reported outcomes; ITT versus per-protocol; power/precision and adverse events.

## Observational study
- Sampling and selection bias; temporality; confounding and its control; measurement validity; missingness; subgroup analyses; association versus causation; sensitivity analyses; generalizability.

## Diagnostic/prognostic study
- Participant spectrum; reference standard; blinding; threshold selection; calibration/discrimination or sensitivity/specificity with uncertainty; prevalence and transportability.

## Qualitative study
- Sampling and researcher positionality; collection and analysis methods; reflexivity; contradictory cases; saturation claims; connection of themes to participant data; transferability.

## Review or meta-analysis
- Question and eligibility; database search and date; screening and duplicates; extraction; risk of bias; model choice, heterogeneity, small-study effects, certainty of evidence and transparency.

## Reporting discipline
- Distinguish absent reporting from definitely absent study procedure.
- Never infer sample size, effect size, p-value, CI, SD, or outcome timing from unsourced context.
- Preserve evidence provenance: page/section, directly quoted passage when needed, and whether the claim is original authors' assertion or your inference.