This slide combines the % change shown
before with the absolute change in PWT, as shown here in the yellow
histograms.
At 90 days, there was a 0.6 minute improvement in the placebo
group, a 1.8 minute improvement in the SINGLE group, and a 1.5 minute
improvement in the DOUBLE group.
The overall ANOVA is intended to detect a
difference among the three groups and showed a p=0.075. Therefore we cannot
formally reject the null hypothesis at the conventional threshold p=0.05.
However, the alternate prespecified analysis uses ANOVA of ranks, which may
be more statistically conservative, and shows a p=0.035, for a difference among
the three groups.
These p values are sufficiently low that pairwise analysis is
appropriate.
Pairwise comparison of SINGLE vs placebo showed p=0.026, while comparison of
DOUBLE vs placebo showed p=0.45.
Therefore in this phase II proof-of-concept trial, we interpret these findings
to show that a single infusion of FGF-2 is associated with a significant improvement
in PWT compared with placebo, and that DOUBLE infusion is not superior to
SINGLE.
The likelihood that we are
wrong, in falsely rejecting the null hypothesis, is somewhere between
3.5%-7.5%.
Please note that this study was powered to detect a 60s increase over placebo in
SINGLE and a 90s increase over placebo in DOUBLE, with the assumption that
DOUBLE was both safe and superior.
DOUBLE was not in fact superior to SINGLE and this may account for the
borderline ANOVA findings (0.075)
We reviewed these data from multiple perspectives, and the statistical
findings are essentially the same regardless central tendency measure selected
(geometric vs arithmetic means).