Donepezil, rivastigmine, galantamine and memantine for ...

18
Clinical effectiveness
particularly important when using subjective
outcome measures.
Seven studies adequately reported the point
estimates and measures of variability for all
outcomes; however, none of these included an
appropriate ITT analysis. All studies state that an
ITT analysis was undertaken; however, this was
defined as being on all participants randomised to
treatment who received at least one dose of
medication and who had baseline and at least one
post-baseline assessment of efficacy (LOCF). In
each case some participants were missing from the
analysis as they did not have at least one postbaseline
assessment, and therefore these studies
were judged to have inadequate ITT analysis.
All but one study 48 gives details of the numbers of
withdrawals from the study; however, four do not
give reasons for withdrawal.
[Commercial/academic confidential information
relating to Seltzer and colleagues 55 removed]
Assessment of effectiveness
Cognitive outcomes
ADAS-cog
Seven parallel RCTs and one crossover RCT
report the ADAS-cog (Homma and colleagues 44
use a Japanese version) and results are given in
Table 5. Results in this table are ordered by the
number of comparison groups and then
alphabetically: the results from the trials with two
comparisons are shown first followed by the trials
with three and four comparisons. Results for the
crossover study were presented in several different
ways, for groups A and B, for all who received
placebo (whether in A or B) and similarly for all
who received donepezil, for those who completed
both parts of the crossover and for whom no data
points were missing. In the current review only
data reported in terms of donepezil treatment
versus placebo treatment (therefore treated as a
parallel comparison) are discussed. The ‘ITT’
(LOCF) results are presented unless not reported
in an individual trial. These conventions will apply
throughout the report. One trial only reported the
treatment difference and the statistical significance
value without any data and has therefore been
omitted from this table. 48 In this and later tables,
values estimated from figures are given in italics.
On the ADAS-cog a negative mean change
indicates a clinical improvement.
Three trials were two-arm comparisons, three were
three-arm trials and one was a four-arm trial.
Changes from baseline scores for each individual
trial are given in Table 5. The summary that
follows will predominantly discuss comparisons
between 5 mg donepezil and placebo and between
10 mg donepezil and placebo, regardless of the
number of arms in the individual trial. The one
study 45 where the intervention dose of donepezil
was 5 mg/day for 28 days followed by 10 mg/day
until study completion will be treated as having a
10 mg/day dose.
Donepezil 5 mg/day versus placebo. Six trials
included an intervention group with a daily dose
of 5 mg donepezil. The mean change from
baseline ADAS-cog score was between –2.43 and
0.2 for the donepezil group and between 0.11 and
3.2 for the placebo groups. Some of this variation
may be explained by differences in the sample
sizes, length of follow-up and study quality.
Sample sizes ranged from 103 in the study by
Greenberg and colleagues 54 to 545 in the two
relevant groups in Burns and colleagues’
comparison. 50 Length of follow-up was 24 weeks in
five studies and 12 weeks in two studies. 52,53 No
studies reported an adequate ITT analysis, and
only Greenberg and colleagues 54 were rated as
having adequate randomisation and concealment
of allocation procedures. Blinding of outcome
assessors was only adequate in two studies 52,54 and
blinding of care provider and participants was
rated as being adequate in only one study. 54
Baseline ADAS-cog scores were similar between
trials. The mean treatment difference in change
from baseline scores between donepezil and
placebo was ~0.90. Overall, all studies found that
ADAS-cog scores were statistically significantly
lower (better) with 5 mg donepezil per day than
placebo.
Three of the six studies provided data (mean
change and standard deviation) that allowed them
to be combined in a meta-analysis (Figure 1).
Pooling the data using a fixed-effect model
showed an overall improvement in ADAS-cog with
5 mg/day donepezil compared with placebo [WMD
–2.51 (95% CI: –3.26 to –1.76)]. Heterogeneity
was not significant at p = 1. No difference was
noted using a random-effects model.
Donepezil 10 mg/day versus placebo. Four trials
included an intervention group with a daily dose
of 10 mg donepezil. The mean change from
baseline ADAS-cog score was between –2.7 and 0.2
for the donepezil group and between 0.4 and 3.2
for the placebo groups. Some of this variation may
be explained by differences in the sample sizes
which range from 66 in the study by Krishnan and