Table 3.

Estimations of sample size per arm for a hypothetical trial to detect 25% rate slowing with 90% power and α = 0.05

Study design	Alzheimer's disease	Mild cognitive impairment
MRI Hippocampal volume
Two scans, 0–6 ma	462	949
Two scans, 0–12 ma	252	698
Three scans, 0–6–12 ma	255	673
Three scans + ApoE4b	196	672
Three scans + MCc	140	383
Three scans + MC + ApoE4d	86	341
ADAS-Coge
Two tests, 0–6 m	745	4663
Two tests, 0–12 m	814	9350
Three tests, 0–6–12 m	569	8354
MMSEe
Two tests, 0–6 m	1280	6300
Two tests, 0–12 m	1083	3900
Three tests, 0–6–12 m	780	3353

a Based on MRI scans at baseline (0 m), 6 (6 m) and/or 12 months (12 m).

b Accounting for ApoE4 profile.

c Using the Markov chain approach (see text for details).

d Using ApoE4 and Markov chain together.

e Using the Markov chain approach did not significantly improve sample size.