exhaustive, because almost all clinical trials are
different. What follows is an attempt to briefly
review the classes of clinical trial design that will
encompass a large majority of studies, and to com-
ment on their economy and end point possibilities.
Parallel-group studies are typically thought of as
the most straightforward design case. In fact, a
bewildering array of variations exists within this
class.
In the simplest case of parallel-group study, a
group of patients presenting sequentially are ran-
domized to one of two equally sized treatment
groups, until a prospectively determined total num-
ber of patients has been recruited. All these patients
are followed for a predetermined period of time, or
until some end point is achieved. The database is
quality assured and locked before the randomiza-
tion code is broken. The patients are then sorted
according to their treatment, the end point mea-
surements are subjected to a statistical test and an
interpretation of the effect (or absence thereof) of
the drug is made. What could possibly go wrong?
The answer is that little can go wrong when there
are ample patients, plenty of drug available, the
choice of dose size has been perfect, the end points
are incontrovertible, the measurements are possible
using a rational or absolute scale, there is ample
toxicological coverage for all the dose sizes
employed and the trialist has an unlimited budget!
This combination of Utopian conditions never exists.
The ascending dose-ranging cohort design is one
variant within the parallel-group class. It is best
suited when there is no cast-iron assurance of
tolerability for all thedose sizes ofinterest. Patients
are randomized in cohorts to either active or pla-
cebo treatment; frequently there are fewer placebo-
treated patients in each cohort.
The objective is to cumulate tolerability experi-
ence as dose size gradually increases. If the treat-
ments in the first cohort prove to be well tolerated,
then the next cohort is randomized in the same way
except that the active-treated patients receive a
larger dose size. Note that this judgment can be
made without breaking the blind. A comparable
number of placebo-treated patients to any single
active-treatment group can be cumulated across
several cohorts, each cohort having fewer placebo-
than active-treated patients. This economizes on
patient numbers in comparison to randomizing
each cohort in a 1:1 fashion, and may also econo-
mize on both drug and patients if two doses are
found to be similarly effective and well tolerated,
albeit not the highest dose that was projected.
Sequential cohorts do not usually economize on
time. Treatment codes can be broken at the end of
each cohort (and not introduce bias into observa-
tions of succeeding cohorts). Sometimes, this can
lead to early closure of the study when the desired
pharmacodynamic effect is observed at a lower dose
than the maximum projected by the study. However,
the deliberations of safety committees at the end of
each cohort can often be time-consuming.
Within-patient dose titration designs may be
conceptualized as the application of an ascending
dose cohort design within a single patient. The
advantages of such designs are when immediate
high-dose therapy is contraindicated for tolerabil-
ity reasons, and when there is likely to be large
variationsbetween patientsin the tolerability and
efficacy of the test drug.
Patients are reviewed during and after comple-
tion of a course of therapy which may include
programmed changes in dose size. If the drug is
well tolerated they may progress to a course of
therapy at higher dose. A prospective limit on
dosing and the number of courses of treatment is
made (e.g.according to toxicologycoverage). Dos-
ing may be curtailed at any time when either there
is unreasonable intolerance of the drug, or when
acceptable efficacy and simultaneous tolerability
have been observed. This is not unlike the approach
to therapy under ordinary clinical circumstances.
For example, patients with epilepsy are often trea-
ted by dose alterations. Another advantage of this
design is that at the end of the study, the range of
tolerated and efficacious doses can be examined
among all treated patients in comparison to demo-
graphic factors, disease subtypes and so on.
The greatest difficulty with ascending-dose,
within-patient designs is usually in treatment
masking. Double-blind requirements have to take
into account a wide variety of dose sizes, and that
contemporaneous placebo formulations will be
needed. Some studies of this type are hybridized
with a crossover strategy (see below). Dose tailing
at the end of the study may be viewed as the same108 CH9 PHASE II AND PHASE III CLINICAL STUDIES