Systematic reviews and meta-analyses

A systematic review gathers every study meeting pre-set criteria and appraises their quality. A meta-analysis statistically combines their results into one estimate. Together they reduce the influence of any single unusual study, which is why they sit near the top of most evidence hierarchies.

A systematic review searches all relevant studies to a stated method; a meta-analysis pools their numbers. Here is why that outranks a single study, and when.

Key takeaways

What makes a review systematic

A narrative review is one expert's selection. A systematic review states its question, search strategy, databases, inclusion and exclusion criteria and appraisal method before searching, so someone else could repeat it and reach the same set of studies. Reporting guidelines such as PRISMA exist so readers can check that this happened.

What a meta-analysis adds

Pooling results across studies increases the effective sample size and narrows the uncertainty around an estimate. The output is usually a summary effect with a confidence interval: if that interval crosses no effect, the pooled evidence does not establish a benefit even when individual studies looked positive.

Heterogeneity: when pooling stops making sense

Combining studies only helps when they ask a similar enough question. Different doses, durations, populations and outcome measures produce heterogeneity, reported with statistics such as I-squared. High heterogeneity means the pooled number hides real differences and should be read as a rough average of unlike things.

Publication bias

Studies with positive results are more likely to be published, published faster, and published in English-language journals. Reviews test for this with funnel plots and similar methods, but they cannot recover studies that were never written up. Where publication bias is likely, the true effect is usually smaller than the pooled estimate.

Garbage in, garbage out

A rigorous review of weak studies is still weak evidence. Good reviews grade the certainty of evidence - often using an approach such as GRADE - and state plainly when the answer is 'we do not know'. A review that concludes 'further high-quality research is needed' is reporting a genuine result.

Frequently asked questions

Is a meta-analysis always better than a single trial?

No. One large, well-conducted, pre-registered trial can be more informative than a meta-analysis of small, poorly controlled studies. Read what went into the pool.

What is a Cochrane review?

A systematic review produced to Cochrane's methodological standards, which are among the most stringent in health research and include explicit handling of bias and conflicts of interest.

What does 'low certainty evidence' mean?

That further research is likely to change the estimate. It is a statement about confidence in the finding, not a statement that the intervention does not work.

Related on Beiing

Sources

  1. Cochrane Handbook for Systematic Reviews of Interventions - assessing certainty of evidence - Cochrane
  2. Page et al. (2021), The PRISMA 2020 statement: an updated guideline for reporting systematic reviews - BMJ / PubMed
  3. PubMed - National Library of Medicine - NIH
  4. National Institutes of Health (NIH) - NIH

Written by Beiing Editorial Team · Editorially reviewed by Research Desk · Published · Last reviewed . Editorial and research review by Beiing writers and researchers. This is not medical or clinical review, and no clinician signs off on our content.

Educational content only - not medical advice. In a crisis, call or text 988 (US) or your local emergency number. See crisis resources.