Assessing biases and errors in systematic reviews and meta-analyses

Systematic reviews and meta-analyses play an important role in evidence-based decision making. However, the validity of findings may be threatened by various sources of bias and error. In this thesis I describe the extent of bias due to selective inclusion of results and its impact on meta-analytic estimates, and the risk of bias due to missing evidence, in systematic reviews of nutrition research. I also collated many errors that might arise during data extraction, statistical analysis and interpretation of results of meta-analyses, which will inform the development of a checklist to detect statistical errors in meta-analyses.