To estimate the time (and corresponding 95% confidence intervals [CIs]) to the CDR‐SB values associated with loss of independence in IADLs or BADLs, we adapted a model for estimation of time savings in AD treatment trials. 30 Specifically, we used the statistical models of the association between time (x‐axis) and CDR‐SB (y‐axis) from our modeling and converted them to estimate CDR‐SB as a function of time with corresponding CIs. For clinical trial data, we used an analogous approach to estimate time to a CDR‐SB cutoff for the placebo and treatment groups based on the published changes in CDR‐SB. 1 , 2
