Biomedical Engineering Reference
In-Depth Information
terminate a study earlier than anticipated and the stopping rules are
generally always made explicit and are not infrequently invoked.
But another more vexing problem arises. What should one say to
patients who show up late in the study, once a trend has become
clear? Can one then assert that the trial is still in equipoise? I don't
see how one can.
Two tactics have been employed to get around this difficulty. The
first is that the doctor treating the patient is blinded to the results of
interim analyses of the study, relying on the study center to monitor
the study and terminate it if necessary. In this way, the doctor would
not know if the patient were indeed being subjected to poorer, though
not “significantly” so, odds. I have a great deal of difficulty with this
“solution”. It seems to me that consciously avoiding learning infor-
mation that, if given to the patient, might change his or her mind
is incompatible with the trust the patient places in the doctor, and
with the patient's legitimate expectation that the doctor be as
knowledgeable as reasonably possible, both in general and with
respect to the study the doctor is proposing.
The second tactic takes advantage of the fact that, in some studies, the
results can only be evaluated long after treatment. All patients
necessary to establish statistical significance may then be entered into
the study before the results begin to be evident. This seems to me to
be more acceptable - although it is still something of a subterfuge,
since accrual into the study could be regulated so that interim results
could be used to modify and perhaps terminate the study.
Cost-benefit trials
There is a class of RCTs that are aimed, not at the issue of whether
one therapy or diagnostic approach is better than another, but on
whether the improvement offered by a new and costly therapy or
procedure is sufficient to be worth the additional cost. In such trials,
there is little doubt that the experimental arm is superior; the trial is
designed to measure just how much better it is.
If one were completely open in seeking the patient's informed consent
to participate in such a trial, one would have to say to the patient “I
want you to take a 50% chance of receiving inferior care so that we
can determine whether society can afford to give the better care to
everyone in the future.” Few patients would agree to do this if it were
presented in such a direct manner. This being the case, I think that,
generally, such trials should not be attempted.
