Abstract:
|
Outcome reporting bias (ORB) and publication bias (PB) are a threat to the validity of network meta-analysis (NMA). In recent years, multivariate meta-analysis (MMA) approaches have been proposed to handle the impact of potential ORB in the pairwise meta-analytic setting. These methods have shown that MMA can reduce bias and increases efficiency of pooled effect sizes. We present Bayesian methods for multivariate NMA (MNMA) that can reduce the effects of ORB and PB on pooled effect sizes. We perform several simulation studies that show MNMA reduces the bias of pooled effect sizes under a variety of outcome missingness scenarios, including missing at random and missing not at random. We apply the approaches to networks in alcohol use and depression.
|