Prediction intervals for random-effects meta-analysis: a confidence distribution approach

TL;DR

This paper introduces a bootstrap-based method for constructing prediction intervals in random-effects meta-analysis, providing more accurate coverage especially with few studies, compared to traditional methods.

Contribution

The paper proposes a novel bootstrap approach for prediction intervals in meta-analysis that improves accuracy over existing methods in small-sample scenarios.

Findings

Bootstrap method achieves coverage close to nominal levels in simulations.

Traditional Higgins-Thompson-Spiegelhalter intervals depend on large-sample assumptions.

Applied method successfully in three real meta-analyses.

Abstract

Prediction intervals are commonly used in meta-analysis with random-effects models. One widely used method, the Higgins-Thompson-Spiegelhalter prediction interval, replaces the heterogeneity parameter with its point estimate, but its validity strongly depends on a large sample approximation. This is a weakness in meta-analyses with few studies. We propose an alternative based on bootstrap and show by simulations that its coverage is close to the nominal level, unlike the Higgins-Thompson-Spiegelhalter method and its extensions. The proposed method was applied in three meta-analyses.