Meta-Analysis and Systematic Review

Definition & place in the evidence hierarchy

• Systematic review — A review using explicit, pre-specified, reproducible methods to identify, select, critically appraise and synthesise all studies addressing a clearly formulated question — the analysis is laid out in a protocol before any results are seen.
• Meta-analysis — The optional statistical step within a systematic review — the quantitative combination of results from two or more separate studies into a single pooled estimate. A systematic review need not contain a meta-analysis.
• vs narrative review — Unlike a traditional/narrative review, the systematic review pre-specifies its question, eligibility criteria, search, appraisal and analysis, minimising the reviewer's selective, post-hoc emphasis on favoured studies. Where studies are too diverse, biased or affected by reporting bias, a narrative (qualitative) synthesis is preferred and a pooled estimate may mislead.
• Four synthesis questions — Any synthesis answers: (1) direction of effect, (2) size of effect, (3) consistency across studies, (4) strength of evidence. Meta-analysis addresses 1–3; question 4 needs judgements on design, risk of bias and certainty (GRADE).
• Apex of the pyramid — A rigorous systematic review / meta-analysis of randomised controlled trials sits at the top of the evidence hierarchy for therapeutic questions — it integrates the totality of unbiased primary evidence. An individual-patient-data (IPD) meta-analysis is the "gold-standard" form. This is the synthesis layer — not primary trial design.