Cumulative meta-analysis: What works

Abstract

To present time-varying evidence, cumulative meta-analysis (CMA) updates results of previous meta-analyses to incorporate new study results. We investigate the properties of CMA, suggest possible improvements and provide the first in-depth simulation study of the use of CMA and CUSUM methods for detection of temporal trends in random-effects meta-analysis. We use the standardized mean difference (SMD) as an effect measure of interest. For CMA, we compare the standard inverse-variance-weighted estimation of the overall effect using REML-based estimation of between-study variance ${\tau }^2$ with the sample-size-weighted estimation of the effect accompanied by Kulinskaya-Dollinger-Bjørkestøl (Biometrics. 2011; 67:203-212) (KDB) estimation of ${\tau }^2$ . For all methods, we consider Type 1 error under no shift and power under a shift in the mean in the random-effects model. To ameliorate the lack of power in CMA, we introduce two-stage CMA, in which ${\tau }^2$ is estimated at Stage 1 (from the first 5-10 studies), and further CMA monitors a target value of effect, keeping the ${\tau }^2$ value fixed. We recommend this two-stage CMA combined with cumulative testing for positive shift in ${\tau }^2$ . In practice, use of CMA requires at least 15-20 studies.

Keywords: CUSUM charts; effective-sample-size weights; inverse-variance weights; power; type 1 error.