First published online: 3 November 2020, Int J Epidemiol 2020;49:2021–29, doi: https://doi.org/10.1093/ije/dyaa156
In the section “Estimating the effects of dynamic baseline interventions,” conditional exchangeability of treatment, specifically independence between the potential outcomes and observed treatment conditional on baseline covariates, is listed as one of the identification conditions needed for the authors’ proposed approach to be valid. This condition was incorrect.
The section has been corrected, and an additional file exploring this condition in more detail has been added to the Supplementary data.