“…To evaluate changes on relevant outcome measurements from baseline to discharge and discharge to follow‐up, we used multilevel models, conducted in R using the nlme package (Pinheiro, Bates, DebRoy, Sarkar, & R Core Team, ). To handle high levels of missing data, we used full information maximum likelihood estimation, which is considered a robust method for generation of model estimates with missing data (Allison, ) and generates estimates similar to multiple imputation methods (Ferro, ; Grund, Lüdtke, & Robitzsch, ; Peyre, Leplege, & Coste, ). Multilevel models included repeated measurements of dependent variables (EDE‐Q, binge eating and purging frequency, %EBW, BDI‐II, STAI‐Trait) nested within participants.…”