Objective: Attrition is a common problem in health behaviour change (HBC) trials. When the degree of attrition differs between treatment conditions, then this is called differential attrition and is regarded as a major threat to internal validity. The primary research question of this study was: how often and to what degree does differential attrition occur in HBC trials?
Design: A systematic review and meta-analysis of a random selection of HBC trials (k = 60). We meta-analysed the relative attrition rates using a random-effects model and examined the relationship between the relative attrition rates and the potential moderators: the amount of human contact in delivery and the intensity of the intervention/control condition, the type of control condition, and the follow-up intensity and duration.
Main outcome measures: Relative attrition rates.
Results: The average attrition rate was 18% (SD = .15; M = .15) in the intervention and 17% (SD = .13; M = .13) in the control conditions. The estimated average relative attrition rate was 1.10 (95% CI: 1.01-1.20, p = .02), suggesting an overall higher attrition rate of 10% in the intervention conditions. This relative attrition rate was not related to any of the potential moderators.
Conclusion: There is indication of a slightly higher amount of attrition on average in the intervention conditions of HBC trials.
Keywords: RCT; bias; differential attrition; health behaviour change; internal validity.