Corrigendum to: Fusion designs and estimators for treatment effects

In our paper “Fusion designs and estimators for treatment effects,”1 we demonstrate identification for a particular fusion parameter for bridged treatment effects and propose an estimator for that quantity. Though the parameter was formally identified and the estimator grounded in standard statistical arguments, the estimate of the variance, which treated the nuisance parameters as fixed, was only proposed based on the form of other similar estimators and explored via simulation. Though it performed well in simulations, recent work by Sarah Reifeis and Michael Hudgens2 suggests our proposed variance estimate is not consistent and is not guaranteed to be conservative, as is the case with standard inverse probability of treatment and censoring weighted estimators. While the proposed estimator is expected to be consistent for the point estimate, we no longer recommend the use of the proposed variance estimator. Instead, we recommend that users of the estimator employ the bootstrap with nuisance parameter estimation included in each bootstrap replicate to properly account for the estimation of the unknown nuisance parameters. Data sharing is not applicable to this article as no new data were created or analyzed in this study.