The mean change from baseline in PedMIDAS score over time for the three groups was assessed using a linear regression model for each of the three pairwise comparisons (Amitriptyline relative to placebo, Topiramate relative to placebo, and Amitriptyline relative to Topiramate), adjusted for the baseline PedMIDAS score. The models, and corresponding odds ratios, were adjusted for the two stratification variables (age and baseline number of headache days).

The mean change from baseline in PedMIDAS score over time for the three groups was assessed using a linear regression model for each of the three pairwise comparisons (Amitriptyline relative to placebo, Topiramate relative to placebo, and Amitriptyline relative to Topiramate), adjusted for the baseline PedMIDAS score. The models, and corresponding odds ratios, were adjusted for the two stratification variables (age and baseline number of headache days).

The mean change from baseline in PedMIDAS score over time for the three groups was assessed using a linear regression model for each of the three pairwise comparisons (Amitriptyline relative to placebo, Topiramate relative to placebo, and Amitriptyline relative to Topiramate), adjusted for the baseline PedMIDAS score. The models, and corresponding odds ratios, were adjusted for the two stratification variables (age and baseline number of headache days).

Another major secondary objective was to determine if the change in absolute headache days from baseline to week 24 differed between treatment groups. This objective was assessed using linear regression models, with three pairwise comparisons between treatment groups (Amitriptyline vs. Placebo, Topiramate vs. Placebo, and Amitriptyline vs. Topiramate). The models, and corresponding odds ratios, were adjusted for the two stratification variables (age and baseline number of headache days).

Each comparison was tested using a Bonferroni corrected significance level of 0.017 (0.05/3).

Another major secondary objective was to determine if the change in absolute headache days from baseline to week 24 differed between treatment groups. This objective was assessed using linear regression models, with three pairwise comparisons between treatment groups (Amitriptyline vs. Placebo, Topiramate vs. Placebo, and Amitriptyline vs. Topiramate). The models, and corresponding odds ratios, were adjusted for the two stratification variables (age and baseline number of headache days).

Each comparison was tested using a Bonferroni corrected significance level of 0.017 (0.05/3).

Another major secondary objective was to determine if the change in absolute headache days from baseline to week 24 differed between treatment groups. This objective was assessed using linear regression models, with three pairwise comparisons between treatment groups (Amitriptyline vs. Placebo, Topiramate vs. Placebo, and Amitriptyline vs. Topiramate). The models, and corresponding odds ratios, were adjusted for the two stratification variables (age and baseline number of headache days).

Each comparison was tested using a Bonferroni corrected significance level of 0.017 (0.05/3).

Tolerability was assessed by calculating the percent of subjects who completed the entire 24 week treatment period and the corresponding 95% confidence intervals for each treatment group. Concerns regarding tolerability would be raised if the upper bound of a confidence interval was less than 65% or if the percent of subjects who completed the entire 24 week treatment period in either of the active treatment groups was significantly less than the placebo group.

Tolerability was assessed by calculating the percent of subjects who completed the entire 24 week treatment period and the corresponding 95% confidence intervals for each treatment group. Concerns regarding tolerability would be raised if the upper bound of a confidence interval was less than 65% or if the percent of subjects who completed the entire 24 week treatment period in either of the active treatment groups was significantly less than the placebo group.

Tolerability was assessed by calculating the percent of subjects who completed the entire 24 week treatment period and the corresponding 95% confidence intervals for each treatment group. Concerns regarding tolerability would be raised if the upper bound of a confidence interval was less than 65% or if the percent of subjects who completed the entire 24 week treatment period in either of the active treatment groups was significantly less than the placebo group.

The occurrence of treatment-related SAE's was assessed in two ways: 1) percentage of subjects who experience any treatment-related SAE in each of the three groups was compared using Fishers exact test; 2) rates of treatment-emergent SAS's across the three groups were compared using a Poisson regression model. However, due to the small number of SAEs that were deemed treatment-emergent (4 in AMI, 1 in TPM, 0 in PBO), the parameter estimates did not converge for the Poisson regression model.

The occurrence of treatment-related SAE's was assessed in two ways: 1) percentage of subjects who experience any treatment-related SAE in each of the three groups was compared using Fishers exact test; 2) rates of treatment-emergent SAS's across the three groups were compared using a Poisson regression model. However, due to the small number of SAEs that were deemed treatment-emergent (4 in AMI, 1 in TPM, 0 in PBO), the parameter estimates did not converge for the Poisson regression model.

The occurrence of treatment-related SAE's was assessed in two ways: 1) percentage of subjects who experience any treatment-related SAE in each of the three groups was compared using Fishers exact test; 2) rates of treatment-emergent SAS's across the three groups were compared using a Poisson regression model. However, due to the small number of SAEs that were deemed treatment-emergent (4 in AMI, 1 in TPM, 0 in PBO), the parameter estimates did not converge for the Poisson regression model.