Company Name
Protocol Nr. 123
The Methods used:
The GLM model used to examine the difference in drug treatments was:
Main variables = subject + treatment + period.
The main end points were the values of dose ratios AUC (area under the curve) and C
max (maximum concentration). Since the area under the curve is zero when the treatment was Placebo there were only three treatments to consider for AUC and C
max. Thus placebo values for AUC and C
max do not exist. Logarithmic transformations were used to reduce the skewness of the distribution for the test of AUC and C
max. This could not be done for the dose ratio parameters as they contain negative values, between zero and greater than negative one. A pair wise comparison of each dose ratio and placebo was performed using Dunnett's one tailed procedure at the alpha level 0.05. Since the dose ratios could not be transformed, the Shapiro-Wilks statistic was used to determine which residual values might not be considered as samples from a normally distributed population.