This chapter is divided into two sections: (1) Bias within single studies and (2) bias between studies in systematic reviews. The general concept of bias is discussed in the beginning. Particular forms of bias like selection, attrition or detection bias are presented together with potential answers to the problem. The association between potential bias and trial quality is illustrated, as well as the impact of realisation versus reporting and the assessment of trial quality. The first part ends with special considerations regarding bias and quality assessment in diagnostic studies. The main part related to bias at the between-study level contains a description of publication bias and related biases, how it happens, how it may be prevented, and how to deal with it. Further biases like those resulting from choice of databases, biased inclusion criteria, differentially delayed publication, publication language and analysis are also discussed.

Introduction

Traditional narrative reviews were frequently criticised for giving a potentially biased view of a topic [1]. Systematic reviews aim at avoiding many of the biases contained in traditional narrative reviews, but nevertheless there is still potential for numerous biases even in elaborate systematic reviews.

Dealing with biases and compiling systematic reviews belong together. A well-known definition of systematic reviews says [2]: the application of scientific strategies that limit bias by the systematic assembly, critical appraisal and synthesis of all relevant studies on a specific topic. Noteworthy this definition already contains the term bias.