Umbrella Review

An updated systematic review re-conducts the original search (in updated databases) and incorporates new primary studies. An umbrella review synthesizes existing systematic reviews without re-searching for primary studies. If an old systematic review exists and many new RCTs have published, updating the review may be preferable. If multiple different systematic reviews exist (with different populations or outcomes), an umbrella review synthesis is appropriate.

This is double-counting bias. If Reviews A and B both included Study X, you cannot simply add their effect estimates together. Options: (1) Use qualitative synthesis and focus on conclusions rather than pooling numbers; (2) Adjust for overlap (e.g., calculate weighted average accounting for shared studies); (3) Perform sensitivity analysis comparing results with and without overlap adjustment. Document overlap explicitly in results.

Include all systematic reviews meeting your definition of a systematic review (explicit search, selection criteria, quality assessment), regardless of AMSTAR-2 rating. However, weight high-quality reviews (AMSTAR-2 High or Moderate) more heavily in conclusions than critically low-quality reviews. Exclude only true narrative reviews or scoping reviews, which lack the rigor of systematic reviews.

If included reviews are homogeneous in population, intervention, comparator, and outcome measurement, quantitative synthesis (pooling effect estimates) may be possible. However, this is risky due to double-counting. If reviews differ meaningfully in scope or conclude differently, use narrative synthesis, creating a summary of findings table that displays each review's quality, population, intervention, outcomes, and main conclusions. Narrative synthesis is safer and often more transparent.