Evidence and Critical Appraisal

Because the hierarchy was built for questions about whether an intervention works, where randomisation is what balances confounding. Other questions have different threats to validity. Prognosis is best answered by a prospective cohort followed from a common early point, diagnostic accuracy by a cross-sectional comparison against a reference standard, and rare or delayed harms often by observational data, because trials are usually too small and too short to detect them.

A systematic review uses a pre-specified search and selection strategy to identify and appraise all relevant studies on a question. A meta-analysis is the statistical pooling of their results, which a systematic review may or may not include. Pooling studies that are too heterogeneous gives a precise-looking estimate of something clinically meaningless.

It analyses participants in the groups to which they were randomised, whatever they actually received or completed. That preserves the balance randomisation created and reflects real-world effectiveness. A naive per protocol analysis can be biased, because adherence is itself related to prognosis and to events after randomisation, and it often overstates benefit.

That the study was uninformative rather than that the intervention does not work. The interval shows the range of effects compatible with the data, so if it spans both meaningful benefit and meaningful harm the study cannot distinguish between them. Reporting it as evidence of no difference confuses absence of evidence with evidence of absence.