Are current standards of reporting quality for clinical trials sufficient in addressing important sources of bias?

摘要

Determining the quality of a randomized clinical trial (RCT) is necessary for decision-makers to determine the believability and applicability of the trial findings. Issues that are likely to affect the utility of RCT evidence include issues of bias, random error and applicability. In this article we focus primarily on issues of bias and examine the evidence for whether reporting methodological items, including allocation concealment, sequence generation, and blinding of participants can be relied upon as evidence of bias. We present the findings of a systematic review of meta-epidemiological studies and a simulation study demonstrating that commonly examined sources of bias likely play little role in treatment exaggeration. We discuss other issues that may additionally influence trial outcomes including sample size, publication bias, and expertise of trialists. We conclude by discussing strategies to moderate the effect of known biases in assessing overall estimates of treatment effects. (C) 2015 Elsevier Inc. All rights reserved.