Ethics of statistical research
Chris Foy
8 July 2008
Gloucestershire R&D Support Unit
South West Research Ethics Committee
Chris.Foy@glos.nhs.uk
Research on human subjects
•Research funder balances costs of
research against benefits
•But what about the inconvenience and
risks to human subjects (participants)?
•There is a well-developed system for
research in the health field (not only for
trials of medicines)
•Other fields: education, criminal justice …
Ethical principles for research
•Non-maleficence
•Beneficence
•Justice
•Autonomy
•Dignity
•Welfare
•Rights
A trial that went wrong
• Eight healthy volunteers recruited for first
human use of a monoclonal antibody
• Two randomly assigned to placebo
• The other six suffered serious adverse
reactions; some were life-threatening
• Investigation found poor practice
• MHRA now has special precautions for
Phase I trials
A study that couldn’t go wrong?
• A student wishes to interview four cardiac
rehabilitation patients about their
experiences
• Relatively harmless
• But what about the pressure to take part?
• And if poor care is revealed?
• And the safety of the student?
Both need ethical review
• But the depth of the review should be
proportionate to the risks and
inconvenience to participants
• The National Research Ethics Service
(formerly COREC) is moving towards this
Research Ethics Committees in UK
• Started in 1960s
• By 1980s, there were over 200, each with
their own application form and working
practices
• Next to impossible for those conducting
UK-wide research; people took short cuts
• 1991 to 2001 – DH moves towards
standardising and simplifying the system
What is the REC for?
A REC “should safeguard the rights, safety
and well-being of all trial subjects. Special
attention should be paid to trials that may
include vulnerable subjects.” (ICH GCP)
How does the REC do this in
practice?
• Committee has to judge the benefit that
could arise from the research against the
risks and inconvenience to participants
• There’s no fixed rule for striking this
balance, hence a committee weighs it up,
not an individual
• For student studies, the benefit may be
educational
What is the REC looking for?
• No-one is obliged to participate in
research – it’s a favour, so …
• How are participants identified,
approached and recruited?
• Is there a clear explanation of what will
happen to them?
• Do they give genuinely informed consent?
• Do they know they can withdraw at any
time without disapproval?
What is the REC looking for?
• What about the risks (side effects) and
inconvenience (time and travel)?
• What about their study data? Who will see
it? How long will it be kept?
• What about tissue samples (including
body fluids)? How long will they be kept?
Could they be used in other studies? Will
DNA be extracted?
• Will GPs be informed?
What is the REC looking for?
• What will happen at the end of the study
(will study drug continue to be available)?
• What about complaints? Does NHS
indemnity apply? Is insurance needed?
• Has the scientific review job been done, to
show that the research is capable of
delivering benefit?
• Where will the study be published? Will
participants get to know the results?
Special care for vulnerable
participants
• Children (Gillick guidelines may apply)
• Adults without capacity to consent to study
(mental illness, learning difficulty, or
unconscious)
• Those with a terminal illness
• Prisoners
• Dependent on the investigator (e.g. nursing
home residents, students)
• Those whose English isn’t up to informed
consent
Who are the members?
• Maximum of 18 members
• One-third of these are “lay”
• The rest are “expert” – doctors, nurses,
allied professionals, statisticians and
pharmacists
• Open recruitment process
The Committee’s options
• Favourable opinion
• Provisional opinion with a set of written
questions
• No opinion, while the Committee consults
a specialist referee (rare)
• Unfavourable opinion
• All to be done in 60 days
Randomised controlled trials
•First reported trial 1948
•Tuberculosis: streptomycin and bed-rest
vs. bed-rest only
•RCTs called “clinical trials” when therapy
is tested
•Don’t have to be of therapy: e.g. housing
improvements and health
Randomised controlled trial
new treatment
group 1
Outcome
group 2
Outcome
population
control treatment
Why a control group?
•Optimistic about new treatment
•But need to be sure that benefits
wouldn’t have happened anyway
•What is additional benefit of new
treatment against best existing
treatment?
•So compare some subjects who get new
with some who get existing
Why randomise?
• Want to compare the effect of the
treatment and control
• Need to remove other influences
• These include patient and clinician
preference
• So don’t let patient or clinician choose
• Choose by predetermined rule - “random”
Ethics of random allocation
•To a patient unused to the idea, having
treatment chosen by tossing a coin
sounds unnerving, even callous
•Careful explanation is called for
Sample wording for information
sheet
Sometimes because we do not know which
way of treating patients is best, we need to
make comparisons. Patients will be put into
groups and compared. The groups are
selected by a computer which has no
information about the individual, i.e. by
chance. Patients in each group then have a
different treatment and these are compared.
In this study, you will have a 50:50 chance
of receiving X or Y.
Blinding
•In a drug trial, clinician or subject report
may be biased by knowledge of
allocation
•Can avoid this if subject does not know
the allocation (single-blind)
•If clinician does not know either, this is
called double-blind
•Requires identical-looking placebos
Ethics of blinding
•Needs explanation
•Need a mechanism for prompt
breaking of the blind in case of an
adverse event
•“This is a double blind trial which means
that neither you nor your doctor will know
which group you are in (although if your
doctor needs to find out they can do so).”
A modern proposal to experiment
without randomising
•Change standard treatment protocol
every six months
•Compare one cohort with the next
•Avoids explaining about randomisation
•But imports other problems: is the
protocol change an undeclared
experiment?
Quasi-experiment
•Subjects fall into two or more groups
•Not allocated by investigator
•Take advantage of changing conditions
•Also called “natural experiment” or
“experiment of opportunity”
John Snow’s quasi-experiment
•“The Great Experiment” 1850s
•London supplied by two water companies
•One intake in Central London
•Other intake at Thames Ditton
•Compared cholera mortality by source
•Water-borne hypothesis confirmed
•Organism not discovered for 30 years
Surveys
•To gain a broad picture of health,
lifestyle, satisfaction with services etc.
•Usually postal
•Looking for response rate 60%+
•Layout and content are important
•Need reminders
Postal survey problems
•Burden on respondents
•One, two or more reminders?
•Should replies be confidential (numbered,
for reminders and data linkage)?
•Or should they be anonymous (especially
where subject is sensitive)?
Interview survey problems
•Time taken – payments may be offered
•Possibility for upset
•For home interview, safety of researcher
•Stigma – “I saw the bedwetting interview
man on your doorstep”
Cohort study
•Recruit a group and follow them through
time
•Looking for association between factors at
baseline and disease later
•Follow-up period may be years or even
decades
•Needs lots of contact to keep cohort
together
Case-control study
•Looking for association between disease
and factor
•Used where disease is rare: survey
wouldn’t find enough cases
•Choose one or more controls per case
•Typically controls are matched on age
and sex
Case-control study
•Compare factors between cases and
matched controls to estimate relative risk
•Can be difficult to explain to controls why
they are in study
•Example: survey of parents of children
who had meningitis, and control parents
Sample size in quantitative studies
•A sample size calculation should be
done, with reasonable estimates of the
benefit sought
•Too small a study – unethical because it
exposes participants to burden and
hazard for no scientific benefit
•Too large a study – unethical because
wasteful of participants
Qualitative studies
•Used to gather a lot of information
from relatively few respondents
•Don’t use random sampling
•Therefore no claim to be
representative
•Doesn’t limit respondents to
investigator’s agenda
•Not the “soft option” - analysis is hard
work
Publication ethics
•Ethical duty to publish, or the study
might as well not have been done
•Publication needs as much care as
lab or field procedure
•There has been little training for
authors, editors, peer reviewers
www.equator-network.org
Finally
•Research involves a degree of risk
•Jenner wouldn’t have discovered
smallpox vaccination without risk
•RECs ensure that risks are reasonable
and subjects are aware of them - not that
risks are nil
•Plenty still to discover