In their comments about our recent article,1 Dr Fan and colleagues question the use of a completers-only vs an intention-to-treat analysis. A strict definition of the intention-to-treat principle in clinical trials states that all randomized participants are included in the statistical analyses and analyzed according to the treatment group to which they were randomly assigned, irrespective of their treatment and adherence to the protocol. Because some participants did not attend the follow-up visits (6, 12, or 18 months postrandomization), our analyses could not be defined as intention to treat. All other components of the intention-to-treat definition were upheld, and participants with incomplete follow-up data (ie, missing at only 1 or 2 follow-up visits) were considered completers and their outcomes were included in the primary analysis. Furthermore, a prespecified sensitivity analysis using multiple imputation for all randomized participants was included in the article's supplemental information and resulted in similar between-group differences in WOMAC pain scores. Both were statistically significant, and the differences between groups were of uncertain clinical importance. Our approach aligns with guidance published elsewhere.