Disagreements in statistical inferences from the random-effects meta-analyses model based on different between-study variance estimators
*Kristian Thorlund, Copenhagen trial
Keywords: meta-analysis, between-study variance estimator, statistical inferences
Previous studies found inconsistencies between available between-study variance estimators in meta-analyses, but implications to meta-analytical inference in practice are limited since they only measure bias and mean square errors of the estimators. Meta-analytical inferences are chiefly based on treatment effect estimates, confidence intervals, and p-values. We compared meta-analytical inferences from three selected estimators by re-analyzing 297 heterogeneous meta-analyses. Disagreements in estimates of treatment effects and confidence intervals were analyzed pair wise using mixed model linear regression. Disagreements in p-values were determined with respect to the 5% and 1% threshold. Large systematic disagreements in treatment estimates and standard errors were found. P-values were similar. The findings provide useful guidelines for interpretation of sensitivity analysis.