Equipoise is something we don’t discuss enough when it comes to
designing clinical trials. Fries and Krishhnan
quoting Djulbegovic et al define
equipoise as: participants will not suffer relative harm from random
assignment to a particular treatment arm; the results of a study cannot
be predicted consistently in advance; and over a number of RCTs those
proving or failing to prove an hypothesis will be approximately equal in
number. For the majority of randomized controlled trials (RCTs)
performed this is not true because we lack information to make such a
judgment. If strictly applied in this manner, the application of
equipoise would mean that little controlled medical research would be
carried out. Yet, the literature is replete with instances of trials in
which at some point we have a pretty good idea that equipoise is being
badly violated but the trials are allowed to go to completion. This is
not good for the patient.
Most of the time the equipoise we are
discussing is the perspective derived from the clinical community. Under
the best circumstances, patients are asked to enroll based upon the
pooled expectation for the RCT arms and not upon the value of any single
arm (Fig 1).
Equipoise in Clinical Trials
Fig
1: Presentation of a randomized trial protocol for consideration by a
patient. This presents an idealized sequence of invitation, factual
evaluation, ethical valuation, decision, and randomization. Note that
factual evaluation contrasts benefits and risks of usual care versus the
expected benefits and risks of the trial after pooling all arms, and
that the decision point always comes before randomization and hence is
independent of the relative expectations for the different arms of the
trial. Figure courtesy of James F Fries and Eswar Krishnan (Arthritis
Res Ther 2004, 6:R250-R255).
Equipoise in Clinical Trials
The authors go on to argue that “The equipoise principle is replaced here by a new ethical standard
of reasonable ‘positive expected value’ with a higher standard in
protection of the participant. Unlike equipoise, this standard allows
placebo-controlled RCTs with predictable results, where the net effect
over all participants is expected to be a benefit. It does away with the
charade of pretending not to know the likely outcome in advance.” In
other words, the authors justify the abandonment of equipoise since it
is not doing the patient any good and replacing it with a better
alternative.
This is all well and good. However, for some trials
positive expected value can be difficult to calculate. Consider a wound
care RCT in which subjects get randomized to standard of care (SOC) or
SOC plus treatment X. The primary endpoint of the trial is time to heal.
Having studied many such trials in my time it is impossible to say how
long with any certainty it will take to heal a given set of patients.
The assumption is that no matter how long it takes, addition of
treatment X over a specified period of time will shorten the time by
some factor—say 10 or 20%. This statement conveniently ignores the
possibility that many wounds won’t get healed during the trial’s
timeframe. The best we might come up with is that if SOC heals 30% of
wounds over 12 weeks, then we guess that adding treatment X will
increase this percentage to 50%. So, the pooled expected value would be a
10% increase in having a wound healed during the trial’s timeframe
([50-30]/2).
There’s another way to capture the equipoise issue
through trial design. This relatively new trial design also addresses
enrolment/patient retaining issues, a perpetual bane: The way it works
is that patients can either choose to be assigned a random allocation to
intervention + SOC or SOC alone, or select the arm they most want to be
in based on their understanding of risk from investigator input. The
final distribution to the study arms objectively measures patient
equipoise (Omel J & Schwartz K, ASCO Post 2014;5(9)
A Proposal for Patient-Selected Controlled Trials: Good Science and Good Medicine.