Catalogue of bias: attrition bias

that describes attrition bias and outlines its potential impact on research studies and the preventive steps to minimise its risk. Attrition bias is a type of selection bias due to systematic differences between study groups in the number and the way participants are lost from a study. Differences between people who leave a study and those who continue, particularly between study groups, can be the reason for any observed effect and not the intervention itself. Associations for mortality in trials of tranexamic acid and upper gastrointestinal bleeding were no longer apparent after studies with high or unclear risk of attrition bias were removed. Over-recruitment can help prevent important attrition bias. Sampling weights and tailored replenishment samples can help to compensate for the effects of attrition bias when present.