the length of the study. Based on other studies, 34 cognition declines in a “waterfall” shape, suggesting that early reduction of decline may have even greater benefits at later time points. Third, we did not directly investigate the loss of function in ADLs as a function of time but rather related ADLs to CDR‐SB and modeled time until a CDR‐SB value associated with loss of function. Although this enabled us to model treatment effects that were reported in CDR‐SB, this is an indirect analysis. Future studies should explore direct associations between time and loss of ADLs and confirm estimated inflection points in different samples to ensure robustness.
