Statistical analysis
The data were primarily analyzed in the
intention-to-treat population, consisting
of all randomized patients who received
at least one dose of study drug and had at
least one follow-up assessment using the
last-observation-carried-forward approach.
The analysis of all longitudinal
data (such as weight, vital signs, efficacy,
etc.) was performed using ANCOVA,
with the screening/baseline values as the
covariate. The 95% CIs for treatment differences
were estimated using LSMeans.
Interaction effect between treatment and
specific subgroups was also defined in the
model. The analysis of primary end point,
the treatment success rate, was performed
with the Cochran-Mantel-Haenszel test,
with all the stratification variables included
one at the time. The analysis of the
new cases of diabetes was performed using
the hazard ratio. The 0.05 level of significance
was used for all analyses.