The poster indicates that MoCa tests are to be performed annually on folks in the cognition experiment. As indicated before, I do not think the pre-specified test to meaure changes in MoCa scores applied only to the impaired patients - instead I think the pre-specified group was patients 70 or older, regardless of their initial MoCa score.
That notwithstanding, the sample size of impaired patients (n=226, 123 per arm ) seems large enough to analyze independently from the normal cognition older folks (post-hoc) given the high level of power to detect differences in MoCa between groups calculated from the preliminary data on just 108 patients. It is reasonable to assume that the change in MoCa scores in the preliminary assessment was primarily based on the first annual test of patients involved. For the final analysis, more time will have elapsed (up to 3 years and 3 MoCa re-tests for some patients). A much larger change in MoCa is likely to have occurred after 3 years than after 1 year.