Donepezil, rivastigmine, galantamine and memantine for ...

10
Methods
applied by one reviewer and checked by a second
reviewer. Any disagreements were resolved
through discussion. The assessment of the quality
of included trials was limited to published data
only (see discussion, strengths and limitations of
the review).
A number of ‘rules’ for describing the quality of
included trials were prespecified in the current
review to supplement those in Appendix 5; these
are the interpretation of the current review only
and include the following:
Adequate descriptions:
● Randomisation: randomisation by computerised
randomisation schedule whether or not on-site,
use of random tables.
● Allocation concealment: states that
randomisation schedule was concealed from all
personnel, and/or describes how this was
undertaken, and/or randomisation was by
computerised schedule off-site.
● Blinding of care provider (clinician) and
patient: states medications identical in
appearance/matched.
● Blinding of assessor: when stated that testing
undertaken by an assessor blinded to treatment
status or that investigators remained blind to
the treatment group.
● Eligibility: if prestated.
● Reporting outcomes: if mean ± standard error
of the mean (SEM)/standard deviation
(SD)/confidence interval (CI) and others given
for all outcomes.
● Intention-to-treat (ITT): if all patients who were
randomised regardless as to whether they had
any outcome assessment are included in the
analysis.
● Withdrawals: if states numbers and provides
reasons for withdrawal.
Unknown descriptions:
● Randomisation: when the term randomisation
is the only description.
● Allocation concealment: when no description is
given. Unable to assume that using a
computerised randomisation schedule equals
adequate allocation concealment unless stated
that undertaken off-site as it could have been a
printed list.
● Blinding of care provider and patient: when no
descriptions are given.
● Blinding of assessor: when no descriptions are
given.
● Eligibility: if not prestated.
● Baseline: no baseline characteristics given
and/or states that baselines were similar without
giving data, and/or baselines given only for a
subset.
● Reporting outcomes: not a response used.
● ITT: not a response used.
● Withdrawals: no details given anywhere of
numbers withdrawing, including the Ns in the
tables.
Inadequate descriptions:
● Randomisation: states randomised but method
clearly is not (chronological order, case
numbers, etc.)
● Allocation concealment: relates to method of
randomisation above (so if method is
consecutive patients, etc., then inadequate); also,
if states allocation not concealed or the method
used suggests that concealment is not likely.
● Blinding of care provider and patient: states
open study, or that placebo completely different.
● Blinding of assessor: when paper states that
assessor was not blinded.
● Eligibility: if not prestated.
● Reporting outcomes: when no measures of
variance are provided for any of the outcomes
reported.
● ITT: when does not mention ITT/when states
used ITT but method is incorrect, such as last
observation carried forward (LOCF) where
some patients withdrew before any
‘observations’ were made.
● Withdrawals: where numbers clearly do not add
up, or where numbers are only given for the
total group not numbers for each arm.
Partial descriptions:
● Randomisation: where states envelopes were
used but no further details reported.
● Allocation concealment: not a response used.
● Blinding of care provider and patient: just uses
the term double-blind but no further
description.
● Blinding of assessor: when blinding of outcome
assessor to one outcome, such as adverse events,
but does not make it entirely clear whether
blinded to treatment group for all outcomes.
● Eligibility: if only minimal data presented or
just the total group data are presented.
● Reporting outcomes: when measures of variance
are provided for some outcomes but not all
outcomes.
● ITT: not a response used.
● Withdrawals: where numbers are given but no
reasons or numbers and reasons given but
where some still appear to be unaccounted for.
Reported descriptions:
● Only available for baseline characteristics.