Systematic Reviews & Meta Analysis

Summary

Places systematic reviews and meta-analyses at the apex of the evidence pyramid, then contrasts that knowledge-layer view with the DIKW hierarchy — where wisdom, the contextual and value-laden decision, remains a separate step. Walks through the seven Sargeant steps of a systematic review and the supporting infrastructure (PROSPERO, PRISMA, GRADE, RoB 2, ROBINS-I, AMSTAR 2), then turns to the central design choice in meta-analysis between fixed- and random-effects models. Covers forest plots and the heterogeneity statistics that interrogate the pooled estimate — Cochran's Q, I², τ² — and the four investigations (subgroup, stratified, Galbraith plot, meta-regression). Closes with three threats that survive correct technique: publication bias (funnel plots, Begg's and Egger's tests, trim-and-fill), influential studies (leave-one-out diagnostics), and outcome-scale issues (standardised mean differences, combining binary and continuous outcomes). Note: audio companion not yet recorded.