could, therefore, be at the front end – during the
trial planning – rather than the back end – at the
analysis phase.
The study protocol and case
report forms (CRF)
A clinical trial is a complex scientific undertaking
that requires the collaboration of many people: clin-
ical investigators, subjects, study coordinators, data
managers, statisticians, programmers and many
more. It is, therefore, of critical importance that a
study plan, procedures and conventions will be laid
out clearly in advance in a document so that all the
participantsinthisjourney will followthesame road
map. The study protocol is this road map. Like any
good road map, the study protocol must be very
clear about the ultimate goal and direction of the
journey – the study objectives. Often, the clarity of
the study objectives in the protocol determines the
coherence of the rest of the protocol. Clearly and
specifically stated objectives willhelpidentifywhen
the primary measures of efficacy, for example, are
inadequate, or when the design is flawed, or when
superfluous data, which will not contribute anything
to answer the questions posed in the objectives, are
going to be collected.
The creation of the study protocol is a multi-
disciplinary and collaborative effort. Every aspect
of the protocol impacts all other aspects. A medical
procedure may impact the response of subjects to
treatment, their compliance or other important
aspect that ultimately will impact the data and the
conclusions that can be drawn from them. For this
reason, every member of the study design team
must assume responsibility for the entire protocol.
The statistician may be responsible directly for
writing the statistical design considerations and
the analysis plan, yet his or her involvement in all
other aspects of the design that feed into it are
equally important.
The CRF is the data collection tool for the clin-
ical trial. Often, the design of the CRF is viewed as
a technical task auxiliary to the trial, and the sta-
tistician may not see it until the trial is ongoing and
the data start coming in for processing. The CRF
design is an important activity that can make a
difference in the quality of the data obtained in
the trial. It should be viewed as an integral part
of the protocol development process, and input
from the clinician, the clinical monitor, and the
statistician must be obtained. The CRF is a multi-
purpose instrument. It serves the investigator as the
tool for recording the data obtained in the trial,
must facilitate the review of the data by the clinical
monitor, and is the document used by the data
manager to build the database for statistical analy-
sis. The organization and structure of the CRF, the
way the questions are phrased, the use of codes, all
impact the data quality.
Since the early 1980s, technology has been
available for electronic transfer of the data from
the investigational sites to the data center. How-
ever, this technology had very limited used until
recently. Electronic data collection (EDC) systems
have become very powerful and attractive in the
last few years due to the maturity and growing
power and speed of the Internet. The use of EDC
systems can speed up the data processing and thus
ultimately reduce the entire new drug development
process. The procedures involved in the use ECD
systems are still evolving, but one can predict with
certainty that these methods will become the domi-
nant mode of data acquisition in the not too far
future.Analysis and reporting
The analysis of the data at the end of the trial is, of
course, the statistician’s domain. A successful ana-
lysis is one that reaches unambiguous conclusions,
not necessarily the ones the clinical researcher is
hoping for. As we emphasized earlier, the success
of the analysis depends entirely on the way the trial
was conducted and monitored, and theway the data
were generated and collected. The statistician’s
role is to utilize the appropriate tools designed to
most effectively extract the information from the
data. The analysis tools do not create information.
We emphasized the need to prepare for the analysis
at the design stage. It is also important to think one
step ahead and consider the need to analyze the342 CH25 STATISTICAL PRINCIPLES AND APPLICATION IN BIOPHARMACEUTICAL RESEARCH