What Is Publication Bias?

Publication bias occurs when the results of studies influence whether and how quickly they are published, so the published literature misrepresents the evidence. Studies with significant or favourable findings are more likely to be published, cited and found.

Related reporting biases

• Time-lag bias: positive trials publish faster.
• Outcome reporting bias: only favourable outcomes are reported.
• Language and location bias: some studies are harder to find.
• Citation bias: positive results are cited more often.

Why it matters

A systematic review that includes only published studies may overestimate benefits and underestimate harms.

Detecting and limiting it

• Search widely: trial registries, conference abstracts, regulatory documents, preprints and contacting investigators.
• Funnel plots can show asymmetry, which may indicate small-study effects (see How to Interpret a Funnel Plot).
• Statistical tests (for example Egger's test) have limited power with fewer than about 10 studies.
• Compare registered outcomes with published outcomes.
• Sensitivity analyses such as excluding small studies.

Authors' and journals' role

Register trials in advance, publish all results, share data, and encourage publication of null findings.

Illustrative example

Invented for teaching. A review of an antidepressant finds 8 published positive trials. A registry search reveals 4 additional trials with no publication, mostly with null results. Including them reduces the pooled effect by about a third (invented figure for illustration), showing how missing studies can inflate benefit.

What authors of reviews can do in practice

Search trial registries and ask whether registered trials were published; compare outcomes in registrations with those in papers; contact corresponding authors for unpublished data; and report how many registered trials were found without results. State the limits of these efforts in the discussion. Where publication bias is suspected, downgrade the certainty of evidence rather than ignoring the concern.

Common pitfalls

• Equating asymmetry with publication bias (there are other causes).
• Searching only PubMed.

Key takeaways

• Missing results distort evidence.
• Use registries and broad searches; interpret funnel plots cautiously.

This page is general educational information about research methods. It is not medical, legal or regulatory advice; follow your institution's ethics board and your target journal's instructions.