And
Any interim analysis that is not planned appropri-
ately (with or without the consequences of stopping
the trial early) may flaw the results of a trial and
possibly weaken confidence in the conclusions
drawn. Therefore, such analyses should be avoided.
If unplanned interim analysis is conducted, the clin-
ical study report should explain why it was necessary
and the degree to which blindness had to be broken,
and provide an assessment of the potential magnitude
of bias introduced and the impact on the interpreta-
tion of the results. (ICH, E9, 4.5)25.12 Issues in data analysis
Clinical trials present unique problems during the
analysis phase that other experiments do not. The
inherent complexity of the clinical trial is com-
pounded by the fact that it uses human subjects,
and therefore it is governed by a set of ethical rules,
paramount of which is the voluntary and informed
participation of the subjects in the study. Subjects
are required to sign an informed consent form prior
to their enrollment in the study in which they
confirm their understanding of the trial procedures,
the potential risks and benefits, and state their
voluntary agreement to participate. Notwithstand-
ing the informed consent form, subjects can at all
times exercise their free will and choose to termi-
nate their participation, refuse to undergo a proce-
dure, skip avisit or violate anyof the study protocol
procedures without penalty. The result is that clin-
ical trials rarely are conducted exactly as planned.
Some of the issues resulting from this are discussed
below.
Noncompliance, dropouts and missing
data
Noncompliance
In testing the efficacy of a new drug, or studying a
dose–response relationship, it is of critical impor-
tance that subjects take their medication as pre-
scribed in the protocol. Most drugs exhibit a direct
dose–response relationship in terms of the drugs
efficacy and safety. Noncompliance with respect to
the schedule and dose of the study medication may
have serious impact on the researcher’s ability to
determine the recommended dose or even to show
efficacy. When subjects under-dose themselves,
the drug efficacy may be missed and the true
adverse event pattern of the drug may be under-
estimated. Clinical researchers always try to build
in mechanisms into the trial’s procedures designed
to maximize compliance. However, it is not
uncommon that despite such efforts, some subjects
will miss some doses.
It is impossible to adjust for noncompliance at
the analysis phase without making assumptions
about the dose–response relationship, which is
often not well understood and might vary greatly
from one subject to another. It is always important
to assess the level of compliance at the end of the
trial so that one might gain some appreciation,
qualitative and incomplete as it may be, of what
one should expect when the drug is taken as pre-
scribed.
Another type of noncompliance is the subjects’
adherence to the protocol procedures and sche-
dules. It is the role of the investigator to make
sure that the protocol is adhered to. Lack of adher-
ence to the protocol complicates the analysis and
may make the result difficult to interpret.DropoutsSubjects may drop out of the trial for a variety of
reasons. Some could be unrelated to the trial such
as relocation, but others, such as experiencing
adverse events, the perception of no efficacy or
perception of well-being, could be strongly corre-
lated with the study drug effect. The result is that
some subjects will have no data to evaluate from
some time point onwards. When the reason for
dropping out are treatment related, the patterns of
dropouts will be typically different between the
different treatments, and ignoring the missing data
will introduce bias into the analysis. There are a
number of methods for handling dropouts, none
of which is entirely satisfactory. One common way
is to use the Last-Observation-Carried-Forward25.12 ISSUES IN DATA ANALYSIS 339