Transcript Ebola, study design, and Ethics - Global Forum on Bioethics in

Ebola, study design, and Ethics
Nancy E. Kass, ScD
Berman Institute of Bioethics and
Bloomberg School of Public Health
Outline for today
• Ethical tension in “standard” randomized,
placebo-controlled trials
• How this is heightened in Ebola context
• What is an adaptive trial?
• Why is this any different, ethically?
• And what else should we thinking about?
Ethical tensions in RCTs, generally
• Tensions between a doctor’s duties to
the individual patient vs. need to
improve care for future patients
• Between deontological duties vs.
utilitarian commitments
Ethics “solution”: Equipoise
• If we don’t “know” if the experimental
treatment is actually better– if there are no
data to support that one arm is better than the
other-- then we have equipoise
• Yet Problems can Remain:
– False dichotomy --Knowledge is not yes/no- it’s on a
spectrum
– Amount of certainty required for widespread
treatment recommendations may be different from
uncertainties that doctor or patient might accept
Debates for Ebola treatment trials
• Point of View #1:
• Ethically, how can we deny potentially lifesaving
treatment to people who are dying– even
acknowledging some uncertainty?
• Point of View #2:
– Especially when people are dying in vast numbers,
there is an ethical duty to learn what works;
– And there is no ethical violation to use placebos if
we don’t know whether the intervention works
Impact of Children’s Oncology Group: decline in
mortality from pediatric cancer 1975-2000
(thanks to Steve Joffe for slide)
Additional context for treatment
trials
ETHICS IS NOT SIMPLY FIGURING
OUT WHICH “SIDE” POSES BETTER
ARGUMENTS
•Is there another option altogether?
Other options? Adaptive trials?
• In an adaptive trial, the study design is
altered…
–while study is being conducted
–Based on what is learned
• In an adaptive trial, the trial’s design
is changed along the way as we learn
new things about the intervention
Many types of adaptive trials
Example #1
• Adaptive approach #1:
– Multi-arm randomized trial
– Patients randomized to one of several different
experimental treatments or to placebo
– Much higher percentage of patients get something
– Study arms are dropped if interim analysis shows
another arm is better
Examples of adaptive trials?
Example #2
• Start giving experimental drug to a smaller number
of patients (e.g., 40 sick patients)
• Record what happens with survival
• If dramatic (“magic bullet”) e.g., 80-100% survival,
– then we know the drug works!
– No need for randomized trial
Ethics and adaptive trials
• If early results suggest intervention is not
dramatically better than historical
experience, unclear whether it is working…
– Then change design: randomize against placebo
– Less troubling ethically to randomize if unclear
whether or how much drug works
• If interim analysis suggests maybe is better,
change design: randomize unequally e.g.,
70:30, etc.
Ethics and adaptive trials
• In all cases, fewer patients assigned to arm
believed “currently” to be the inferior arm
– Meurer argues (JAMA 2012) this will also reduce
therapeutic misconception
• Commitment to learning with rigor sustained
• Does not eliminate ethical tension but
significantly reduces it
Other design considerations?
• Ethically:
• Must provide supportive care-- may be most
important piece of “compassion” = only thing we
know makes a difference in survival
• How to manage in informed consent? Change consent
language with each design change?
• Technically:
• Only works when endpoints are clear (e.g., death) and
occur relatively quickly
• Requires quality/ongoing DSMB
• Requires nimble and sophisticated study team