Metabias

Metabias, commonly written metabias or meta-bias, refers to biases that arise in the evidence base used for a meta-analysis rather than in any single study. These biases can systematically distort effect estimates and lead to incorrect conclusions about an intervention or phenomenon.

Common metabias types include publication bias, where studies with positive or significant results are more likely to be published; language bias, where studies reported in non-English languages are underrepresented; outcome reporting bias within studies, where only favorable outcomes are fully reported; time-lag bias, where results take longer to publish if they are negative or inconclusive; and selective inclusion of data or study design features that influence results.

Detection and assessment: Funnel plots and statistical tests such as Egger's regression or Begg's test can signal small-study effects; trim-and-fill methods estimate the impact of missing studies; selection models and meta-regression can adjust for some biases. Cumulative meta-analysis can reveal time-lag bias. It is important to consider risk of metabias when interpreting results.

Mitigation: Broad search strategies including gray literature; preregistration of protocols; explicit inclusion criteria for outcomes; contacting study authors for missing data; data sharing; conducting sensitivity analyses; using robust meta-analytic methods that account for potential biases.