University of Zurich
Co-Authors: Giuachin Kreiliger, Leonhard Held
Abstract: Estimating publication bias in clinical trials across different specialties in medicine
Clinical trials study the efficacy of drugs and potential side-effects. Evidence across multiple trials can be summarized in meta-analyses and can influence clinical practice and approval of drugs based on evidence. However, publication bias is a major concern and affects the overall estimates of the meta-analyses (Ioannidis & Trikalinos, 2007; Aert et al., 2019). The aim of this study is a extensive evaluation of publication bias in the Cochrane Database (CDSR). We will evaluate meta-analyses from over 7,000 reviews from the CDSR. Data will be combined with additional journal metrics. We will model publication bias and its associations with medical specialty, total sample size, journal rank of studies, number of citations, publication year, and number of studies included. Furthermore, we will investigate the change in the overall effects after adjusting for publication bias. Estimating publication bias is important as it is a main causes of irreproducible research (Bishop, 2019).