Statistical Analysis
Using a power analysis (expected value, ≥1 – β = 0.8),
it was determined that 40 subjects would be sufficient
for a power of ANOVA ≥0.8 to detect a significant
difference (α = 0.05) in between-formulation 90% CI.21
Assuming a drop-out rate of about 15%, 46 healthy
male Chinese volunteers were enrolled in the present
study.
Cmax, AUC0–t, and AUC0–∞ were considered the primary
variables to assess the bioequivalence between the
test and reference formulations. Two-way ANOVA for
a 2 × 2 crossover design was used to assess the effects
of formulation, period, sequence, and subjects on logtransformed
Cmax, AUC0–t, and AUC0–∞.
22 The nonparametric
signed rank test was used to compare the Tmax
for the 2 formulations. P ≤ 0.05 was considered statistically
significant. The ratios of the log-transformed Cmax,
AUC0–t, and AUC0–∞ of parent atorvastatin and orthohydroxy-atorvastatin
for both formulations were calculated,
and 90% CIs were obtained. ANOVA was