From the Guidelines
Attrition bias refers to the loss of participants during a study, which can lead to biased results if the remaining participants are not representative of the original sample.
Key Factors Contributing to Attrition Bias
- Differential attrition: when the loss of participants differs between treatment groups, which can occur due to various reasons such as side effects, lack of improvement, or worsening of symptoms 1
- High attrition rates: can affect statistical power and lower confidence in the findings, as seen in many trials with small sample sizes 1
- Failure to use intention-to-treat (ITT) analyses: can increase the risk of bias due to missing outcome data, as ITT analysis includes all participants in the groups to which they were originally allocated 1
Consequences of Attrition Bias
- Biased results: can lead to erroneous conclusions, especially if the attrition is related to the treatment response or outcome 1
- Underestimation or overestimation of treatment effects: can occur depending on the method used to handle missing data, such as last-observation-carried-forward (LOCF) or complete case analysis 1
Best Practices to Minimize Attrition Bias
- Use of ITT analysis: to include all participants in the analysis, regardless of whether they completed the study or not 1
- Imputation of missing data: using methods that take into account the uncertainty of imputation, such as multiple imputation or mixed model approaches 1
- Reporting of participant flow: to provide transparency on the number of participants who were randomly assigned, received intended treatment, and were analyzed for the primary outcome 1
From the Research
Definition of Attrition Bias
- Attrition bias is a type of selection bias that occurs due to systematic differences between study groups in the number and the way participants are lost from a study 2.
- It is defined as the loss of participants over the course of a study, which can introduce bias and threaten the integrity of a longitudinal research study 3.
Causes and Consequences of Attrition Bias
- Attrition can occur due to various reasons such as death, illness, worsened health, refusal, withdrawal, or lost to follow-up 4.
- 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 2.
- Attrition bias can impact internal and external validity, reliability, and statistical validity of a study, especially when the sample size is small and non-random 3.
Methods to Reduce Attrition Bias
- Over-recruitment can help prevent important attrition bias 2.
- Sampling weights and tailored replenishment samples can help to compensate for the effects of attrition bias when present 2.
- Imputation and weighting methods, such as regression imputation, individual weighting, or multiple imputation, can be used to limit bias due to attrition, but their effectiveness depends on the missing data mechanism 5.
- Accounting for participants' intent to dropout, such as asking participants to rate their intent to attend the next assessment session, can reduce attrition bias under certain conditions 6.
- Strategies such as tracking, bonding, and incentives can be used to minimize attrition 4.