TableĀ 1

Main sources of bias in clinical studies

Type of biasDescription
Selection biasMissing or inadequate randomisation: participant characteristics, including potential confounders, may differ between the experimental arms.
Performance biasMissing or inadequate blinding of outcome assessor: a physician may be induced to detect improvements in treated individuals.
Detection biasMissing or inadequate blinding of participants: an individual may be induced to perceive health improvements if they know they are being treated.
Attrition biasToo many withdrawals, or unbalanced dropout rates among the arms: loss in statistical power, misleading outcome interpretation.
Reporting biasSelective outcome reporting: in general, authors may be induced to select the findings to report (for example, due to presence of a funding institution).