Placebo responders
Jeremy Miles
I am reviewing a report of a RCT, which has eliminated what they call
"placebo responders" prior to randomisation. That is, beofre
randomization, everyone was placed on placebo for one week anyone whose
score on the outcome dropped by 25% (or more) was eliminated from the
trial - this ended up being about 6% of (previously) eligible
participants.
This seems curious to me. I wondered if anyone could see a good reason
for doing it.
Thanks,
Jeremy
--
Jeremy Miles
mailto:jn...@york.ac.uk http://www-users.york.ac.uk/~jnvm1/
Dept of Health Sciences (Area 4), University of York, York, YO10 5DD
Phone: 01904 321375 Mobile: 07941 228018 Fax 01904 321320
Alexandre Santos Aguiar
> This seems curious to me. I wondered if anyone could see a good reason
> for doing it.
Placebo tipically produces "positive responses" to a variable range. It may
reach 20% of responders!
This may hide a small or difficult to assess (as with subjective measurement)
but significant real effect of the tested drug.
The article below is an example.
Grach M, Massalha W, Pud D, Adler R, Eisenberg E.
Can coadministration of oxycodone and morphine produce analgesic synergy in
humans? An experimental cold pain study.
Br J Clin Pharmacol. 2004 Sep;58(3):235-42.
My 2 cents, :-)
Kindest regards.
--
Alexandre Santos Aguiar, MD
- independent consultant for health research -
R Botucatu, 591 cj 81 - 04037-005
São Paulo - SP - Brazil
tel +55-11-9320-2046
fax +55-11-5549-8760
www.spsconsultoria.com
John Whittington
>I am reviewing a report of a RCT, which has eliminated what they call
>"placebo responders" prior to randomisation. That is, beofre
>randomization, everyone was placed on placebo for one week anyone whose
>score on the outcome dropped by 25% (or more) was eliminated from the
>trial - this ended up being about 6% of (previously) eligible participants.
>This seems curious to me. I wondered if anyone could see a good reason
>for doing it.
It's not difficult to see the perceived 'good reason' for doing it, since
it is more likely to produce a 'positive answer' in relation to the
treatment being tested, but it's very questionable whether it is a sensible
thing to do - unless this is a very early study of a new treatment.
If one adopts that procedure, one's results (be they 'positive' or not)
will relate specifically and only to that subgroup of potential users of
the treatment, and therefore would give no useful information on how the
treatment would perform in a group of patients who had not been 'selected'
in that fashion.
I think that such a practice (as well as other 'selective' practices) can
sometimes make sense in very early studies, by reducing the number of
subjects required to determine whether further study/development should be
undertaken - but that's probably one of the few situations in which it does
make a lot of sense. The real world consists of a heterogenous group of
patients who exhibit varying (and unpredictable) degrees of 'placebo
response', and one generally wants to know how well a treatment works in an
unselected group/sample of such patients.
Kind Regards,
John
----------------------------------------------------------------
Dr John Whittington, Voice: +44 (0) 1296 730225
Mediscience Services Fax: +44 (0) 1296 738893
Twyford Manor, Twyford, E-mail: Joh...@mediscience.co.uk
Buckingham MK18 4EL, UK medis...@compuserve.com
----------------------------------------------------------------
Stephen Senn
are a) unethical and b) a waste of time. Here are some references:
Blackwelder, W. C., Hastings, B. K., Lee, M. L., & Deloria, M. A.
(1990), Value of a run-in period in a drug trial during pregnancy
Controlled Clinical Trials, 11, 187.
Brittain, E., & Wittes, J. (1990), The run-in period in clinical
trials. The effect of misclassification on efficiency Controlled
Clinical Trials, 11, 327.
Schechtman, K. B., & Gordon, M. E. (1993), A comprehensive algorithm
for determining whether a run-in strategy will be a cost-effective
design modification in a randomized clinical trial Statistics in
Medicine, 12, 111.
Senn, S. J. (1995), A personal view of some controversies in allocating
treatment to patients in clinical trials [see comments] Statistics in
Medicine, 14, 2661.
Senn, S. J. (1997), Are placebo run ins justified? British Medical
Journal, 314, 1191.
Senn, S. J. (2001), The Misunderstood Placebo Applied Clinical Trials,
10, 40.
Senn, S. J. (2002), Ethical considerations concerning treatment
allocation in drug development trials Statistical Methods in Medical
Research, 11, 403.
Blackwelder et al, state the view that efficiency is likely to be
improved. Brittain and Schechteman tend to show the opposite. I make
the ethical case against.
I hope that this is of some use.
Regards
Stephen
John Whittington
>... Myopinion is that placebo run-ins are a) unethical and b) a waste of
>time.
Quite apart from the question of 'placebo responders', and whether one
should make attempts to 'take them into account' in trials .....
The nature of some outcome measures is such that one necessarily needs to
look at 'changes from baseline' - or, as a second best, at least establish
that all subjects are roughly similar (in relation to that measure) in the
absence of treatment. More generally, information on the 'untreated state'
is often required in order to establish compliance with trial entry criteria.
Although there are clearly ethical issues here, it seems inescapable that,
in such situations, one cannot conduct a satisfactory trial without some
initial untreated 'washout' period for those subjects previously receiving
treatment - and to give placebo (rather than nothing) during that period
would seem (to me) to be the most obvious thing to do.
I would be interested to hear views on that.
Helveticus
patient is given the experimental treatment and what happend when a
patient is given the control tretament. Obviously, to measure this
difference accurately it is desirable to have comparability at baseline
or to be able to allow for any difference in analysis. However, the
difference from baseline does not form part of the definition of the
treatment effect. Possible risk of interactive side-effects excluded,
there is no need for patient to be washed out in starting the trial and
for many conditions it would be ethically unacceptable for them to be
so. Hnece the requirement is simply that patients should be comparable
at baseline and this is achieved just as well by having an active
treatment in the run-in period as by having a placebo in the run in
period. If it is argued that the actual effect of the treatment at
follow-up depends on the treatment given in the run in period, then
there is no such thing as a unique treatment effect anyway, it depends
intimately on patient history and all experimentation is pretyy well
hopeless.
It is best to consider concrete examples. Take the physician's health
study
1. Hennekens CH, Eberlein K. A randomized trial of aspirin and
beta-carotene among U.S. physicians. Prev Med 1985;14(2):165-8.
In this more than 20,000 US doctors were the subjects. They were given
placebo to beta-carotene in the run-in. In so doing the investigators
misled them as to what was happening. This was unethical and it was
also pointless. Since the follow-up was 12 years, during which time
half the subjects continued to receive beta-carotene, there was no
earthly reason why they could not have all been given beta-carotene in
the run-in. This would have the advantage that there would have been no
need to mislead the subjects.
Regards
Stephen
John Whittington
>The treatment effect is the difference between what happens when a
>patient is given the experimental treatment and what happend when a
>patient is given the control tretament. Obviously, to measure this
>difference accurately it is desirable to have comparability at baseline
>or to be able to allow for any difference in analysis. However, the
>difference from baseline does not form part of the definition of the
>treatment effect.
That surely depends upon ones definition of outcome variable? If that
variable is defined as the change in some quantity from a baseline, then
that becomes reflected in the meaning of 'treatment effect' in relation to
that variable.
>Possible risk of interactive side-effects excluded,
>there is no need for patient to be washed out in starting the trial and
>for many conditions it would be ethically unacceptable for them to be
>so. Hnece the requirement is simply that patients should be comparable
>at baseline and this is achieved just as well by having an active
>treatment in the run-in period as by having a placebo in the run in
>period.
I suspect that one of the problems may be that you and I have different
sorts of trial situations (and outcome variables) in mind, and that is
perhaps underlined by the example you quote.
Before I can really continue this discussion .... I'm not totally sure what
you mean by 'active treatment in the run-in period' - specifically whether
you mean a situation in which all subjects are put on the same dose of the
same active medication during a run-in period, whether you mean that
subjects should be kept on their 'normal medication' (whatever its nature
or dose), if any, during a run-in period (or maybe some third
meaning). The latter ('normal medication', if any) is obviously very
messy, essentially precluding establishment of baseline comparability in a
clinically meaningful sense, and also seriously frustrating attempts to
define the patient population and thereby determine compliance (or
otherwise) with study entry criteria. The former (arbitrarily define fixed
'run-in treatment') carries some of the same advantages of a placebo run-in
(with less ethical problems), but still frustrates attempts to define the
patient population (and compliance with entry criteria) in clinically
meaningful terms. For example, it would be less than clinically ideal to
have (in a trial of an antihypertensive drug) entry criteria which required
a blood pressure between P & Q mmHg whilst taking X mg per day of drug
Y". Furthermore, such an approach may well be hazardous, and thereby
potentially at least as ethically questionable as a placebo run-in, since
imposing an arbitrarily fixed dose of a 'run-in treatment' can, in many
situations, be at least as potentially dangerous as treating only with placebo.
Before I attempt to discuss further, I think I need to ascertain which of
the above-mentioned types of 'active treatment run-in' (or some third one,
if that's the case) you had in mind as an alternative to a placebo run-in
period.