Evidence-appraisal glossary
Prediction interval
In a random-effects meta-analysis, a prediction interval gives the range in which the true effect of a new, similar study is expected to fall. It combines the uncertainty in the average effect with the real variation between studies, so it is wider than the confidence interval around the pooled estimate.
Also called: 95% prediction interval.
It is wider than the confidence interval because it adds the between-study variance, and it is increasingly recommended for showing how much effects genuinely differ across settings rather than just how precisely the average was estimated. The caveats are that it needs enough studies to estimate the between-study variance reliably, and that it predicts the effect in a new context rather than bounding where the already-observed study effects landed.
This is a plain-language methodology definition for reading research. It is general education, not medical advice.