Table 2.

Efficiencies in sample size of adaptive crossover trial compared with traditional and parallel trial designs

Study design	Sample size dose selectiona	Sample size POC efficacy	Total sample size	Power
Traditional parallel armb	62	44	106	0.86
Traditional crossover	54	44	106	0.86
Adaptive crossover	60	24	84	0.86

^aPower calculations performed using G*Power were based on an effect size from a minimum clinically significant difference of 2 points on the NPI apathy/indifference domain score (d = 0.7) and power of 0.80 to permit analysis for males and females, with effect size of 0.7 based on pilot study results of a 3-point difference (d_karr = 0.7) [11]. For comparison, sample sizes assuming a smaller effect size are shown for each design

^bTraditional trial design-based on factor design (analysis of covariance). Four groups (three dose schedules + placebo), df = 3, disease severity as covariate, with sample size doubled to permit separate analysis of males and females

Traditional crossover design comprised three groups, df = 2, disease severity as covariate, with sample size doubled to permit separate analysis of males and females