A unified framework for bias assessment in clinical research.

Abstract

Methodological flaws, limitations, and inadequate practices in research are well known and pose threats to the internal validity of any research study. However, there are ways of safeguarding research conduct to reduce the chance of research producing distorted results. Numerous tools now exist to assess the incorporation of such safeguards into primary research studies (also known as quality and/or risk-of-bias assessment). These tools typically include a variety of items that are then checked against those implemented in the study. Despite a lot of research in this area, no comprehensive generic classification of safeguards across study designs exist, although attempts have been made to clarify aspects of this. We review the developments in this area as well as use preliminary data from 100 methodological studies to illustrate our proposed approach. We conclude by proposing a new framework for identifying research studies at risk of being biased and the information in this article will promote a unification of the diverse approaches to facilitating bias assessment in clinical research.