Biomedical Engineering Reference
In-Depth Information
I want to suggest here that, when
patient numbers allow, one should
perform RCTs with one control arm
and two experimental arms featuring
the same therapy but with different
intensities. One then has the situation
(a)
(a)
(a)
experimental
arm, E
experimental
arm, E
experimental
arm, E
control
arm, C
control
arm, C
control
arm, C
depicted in Figure13.1b. If the in-
tensities of the experimental arms
morbidity (%)
morbidity (%)
morbidity (%)
have been judiciously chosen, one
may be able to interpolate (or even,
modestly extrapolate) between the
results of the two experimental arms
(b)
(b)
(b)
E
1
E
1
E
1
E
2
E
2
E
2
to estimate the TCP of the ex-
perimental arm whose intensity is
C
C
C
such that it leads to the same
morbidity as the control arm. (Of
course, this assumes that a linear
interpolation can be made
TCP gain (%)
TCP gain (%)
TCP gain (%)
morbidity (%)
morbidity (%)
morbidity (%)
which,
for only modestly different results, is
probably a reasonable assumption.)
One can then estimate, as depicted in
Figure 13.1b, the increase in TCP
when the levels of morbidity of the
experimental and control arms are the
same.
−
Figure 13.1. Schematic
representation of the results
of: (a) a standard 2-arm
RCT and (b) a 3-arm trial
where different intensities
of the experimental arms
(E1 and E2) are used (see
text).
Such an approach must include an estimate of uncertainties, and the
process of interpolation or, worse, extrapolation, will magnify these.
In consequence, more patients are required for this type of study than
for conventional two-arm studies.
R
ANDOMIZED
C
LINICAL
T
RIALS
:
NON
-
QUANTITATIVE ISSUES
Finally, I would like to make some brief comments regarding the
social aspects of randomized clinical trials. I base these comments on
two well-established principles, a commitment to which is, I believe,
an important precondition for conducting an RCT and for
recommending a patient to enter one.
Equipoise
A critical principle in justifying randomization is
that one can assure an eligible patient that, given the state of
knowledge at the time, it is truly a toss-up as to which arm will be
found to be superior, if either. This condition is known as
