Data Ethics

1
Data Ethics

• Chapter 7

2
Are sham surgeries ethical?

• Weve just learned that randomized, double-blind,
  placebo-controlled experiments are the gold
  standard for evaluating new interventions
• And, we heard last time that the FDA is required
  to use this method to demonstrate that new drugs
  are safe and effective
• What about surgeries?
• It turns out that only 7 of studies on surgeries
  use a randomized, comparative method

• Weve just learned that randomized, double-blind,
  placebo-controlled experiments are the gold
  standard for evaluating new interventions
• And, we heard last time that the FDA is required
  to use this method to demonstrate that new drugs
  are safe and effective
• What about surgeries?
• It turns out that only 7 of studies on surgeries
  use a randomized, comparative method

3
Are sham surgeries ethical?

• Given this, how do we know surgeries are
  effective?
• The placebo effect works just as well with a
  surgery
• Heres an example there is a new and promising
  treatment for Parkinsons Disease that involves
  implanting new cells in peoples brains
• A randomized, placebo-controlled experiment would
  require a lot of people to have major head
  surgery in order to receive the placebo (a
  solution with no new cells)
• This surgery carries major risks of many kinds

4
Are sham surgeries ethical?

• Is this ethical? Can we subject people to the
  significant risks associated with brain surgery
  just so they can serve as the placebo-control for
  an experiment?
• Thoughtful people disagree on this, but most
  agree on some basic principles to guide ethical
  decisions like this

5
First principles

• The principles of ethics that we will discuss
  concern potentially grey areas where no one
  answer is clearly and absolutely right
• This is not the case for
• Clearcut lies and deception of the sort employed
  by some telemarketers who pose as legitimate
  survey takers in order to elicit information
• Cases in which investigators fabricate fake data
• These are clear cut and obviously wrong
• Heres an example that illustrates the kind of
  question that our ethical principles are designed
  to help answer

6
Example 1 Missing details

• Many scientific journals put strict limits on the
  length of papers they will publish
• This strongly motivates authors to be brief, and
  this provides an opportunity to conveniently omit
  information that may not strongly support the
  conclusion of the paper
• After reviewing more than 4,000 medical papers
  for statistical validity over a ten year period,
  a statistical consultant for the New England
  Journal of Medicine had this to say

7
Example 1 Missing details

• When it came to the statistical review, it was
  often clear that critical information was
  lacking, and the gaps nearly always had the
  practical effect of making the authors
  conclusions look stronger than they should have.
• This is the analysis of an expert statistician
  who was hired to review medical papers over a
  long period of time
• Clearly we need to think about what happening

8
First principles

• The most complex ethical issues arise when people
  are involved in the study
• Experiments, especially medical studies, that
  impose a treatment on people involve the biggest
  ethical challenges
• Experiments can potentially harm the subjects
• Surveys and observational studies still involve
  significant ethical issues, but they are
  generally less problematic
• The law requires that studies funded by the
  Federal Government obey the following ethical
  principles

9
First principles
10
Institutional review boards

• The purpose of and institutional review board
  (IRB) is
• to protect the rights and welfare of human
  subjects (including patients) recruited to
  participate in research activities
• The IRB is not responsible for deciding if the
  proposed study is well-designed or will produce
  valuable new information
• The primary mandate of an IRB is to protect human
  subjects involved with research

11
Institutional review boards

• The activities of an IRB include
• Reviewing planned research protocols can call
  for changes
• Reviews the consent form to be sure that subjects
  are informed of the nature of the study and
  possible risks
• Annual check-up for ongoing research projects

12
Informed consent

• What is informed consent?
• Must be informed
• A description of the study and its purpose
• The potential benefits that might arise
• All of the possible risks and harm that the
  subject may be exposed to
• The potential costs in time, effort and money
• Signifies consent of the subject to participate
  in the research
• Subjects must agree (consent) to participate,
  usually in writing

13
Informed consent

• Sample surveys do not involve physical risks for
  respondents
• What kinds of risks might be involved with a
  sample survey?

14
Example 2 Who can consent?

• What kinds of subjects might not be able to give
  informed consent?
• To provide informed consent, subjects must be
  able to
• Absorb and process information at a level that
  enables decision making about participating in an
  experiment
• make decisions freely without being coerced or
  manipulated
• It used to be common practice to test vaccine on
  prisoners who were paid with good behavior
  credits
• How is this not okay?
• Can a prisoner really say no?

15
Example 2 Who can consent?

• Prisoners and other institutionalized people are
  one category of people who cannot give informed
  consent
• Children are another because children are too
  young to fully grasp the implications of being an
  experimental subject, they cannot give fully
  informed consent
• Parents are usually asked to give consent for
  children
• If parents do not return consent forms sent home
  by a school that wants to enroll children in a
  small sample survey, can the school go ahead
  because the parents have not said no?

16
Example 2 Who can consent?

• What about research on people with mental
  disorders, deep depression for example?
• Is it ever possible to get informed consent in a
  situation like this
• How can we study treatments for mental illness
  ???
• Imagine there is a new treatment for
  unconsciousness that can possible save someone
  who has just suffered a sever head injury from
  slipping into long-lasting coma
• Can we enroll unconscious people into an
  experiment to study this treatment
• What is the alternative?
• These are questions with no clear answers

17
Informed consent

• There are many difficulties and controversies
  surrounding informed consent
• Some researchers view it as a barrier to
  recruiting subjects, which it can be
• As a result, consent documents sometimes fail to
  point out all the risks, or exaggerate the
  possible benefits, or neglect to point out that
  better proven therapies already exit
• Creating a truly informative consent document
  sometimes makes it very long, tedious and scary,
  and this scares off potential subjects

18
Example HIV transmission study

• Study of the effect of recent infection on HIV
  transmission
• Existing sero-surveillance study regularly tests
  everyone in a big population for HIV
• Special tests identify recently infected people
  and not recently infected people
• Recently infected people asked to identify their
  sexual partners so that their serostatus can be
  assessed and monitored
• What does informed consent ask, and from whom
  the primary subject or the possibly nominated
  partner ??

19
Confidentiality

• The first step to conducting research with human
  subjects is to have your institutions IRB review
  and approve your protocol
• Next, the subjects have to provide informed
  consent to participate in the research
• With those two requirements met, the research can
  begin and the data can be collected
• So thats it, our ethical obligations have been
  met!
• Not quite

20
Confidentiality

• Now it becomes important to protect the
  information that has been collected
• The principles here are
• Privacy,
• Information that the subject would not consider
  public must be protected
• Confidentiality
• Is the researchers commitment to ensure that the
  private information describing research subjects
  is not revealed, does not become public
• What can be reported publicly are aggregated
  summaries of individual data or other statistical
  descriptions of the individual data

21
Confidentiality

• This is not the same as anonymity
• Anonymity means that the identities of the
  subjects are completely unknown to everyone
  involved in the study, and this is rarely true or
  desirable
• Prevents re-contacting subjects to improve
  nonresponse or to follow-up
• Prevents communication of study results to
  subjects
• Etc.
• Breaches of confidentiality are serious data
  ethics problems

22
Confidentiality

• One way to ensure confidentiality is to store
  personal identifiers separately from the rest of
  the data and only link the two when absolutely
  necessary
• Unfortunately in the era of computerized
  databases and networks, this is not always enough
• By linking and searching multiple anonymized
  databases, a clever data miner could possibly
  identify and piece together a detailed picture of
  individuals
• This is a concern with all the databases
  maintained by the government and other private
  agencies
• Banks, health care organizations, marketing firms
  etc.

23
Clinical trials

• Clinical trials are experiments on people that
  study the effectiveness of new medical treatments
• Medical treatments can harm as well as help
  people, so the study of new unproven medical
  treatments on people highlights the ethical
  issues involved with experiments on people
• Building on experience prosecuting Nazi war
  criminals during the Nuremburg trials, the 1964
  Helsinki Report of the World Medical Association
  enumerates the key ethical principles
• Link to Helsinki Report

24
Clinical trials

• During World War II the Nazis experimented on
  prisoners, often in very cruel ways that resulted
  in the death of the subjects
• Obviously, this highlights the need to protect
  subjects
• It also reveals a common justification for
  experiments that harm subjects it is surely okay
  to harm a small number of individual people in
  the larger interests of society as a whole?
• Based on these experiences, here are the basic
  guidelines to frame a discussion of the ethics of
  experimenting on people

25
Clinical trials

• Randomized comparative experiments are the only
  way to see the true effects of new treatments.
  Without them risky treatments that are no better
  than placebos could become common.
• Clinical trials produce great benefits, but these
  benefits go to future patients. The trials also
  pose risks, and these are borne by the subjects
  of the trial. We must balance the risk to the
  present subjects with the benefits to future
  patients.
• Both medical ethics and international human
  rights standards indicate the interests of the
  subject must always prevail over the interests of
  science or society.

26
Clinical trials

• We talked last time about the Tuskegee study
  which is a good contemporary example of what
  happens when people ignore these principles and
  act in the interest of people other than the
  subjects
• Because the interests of the subject must
  prevail, new medical treatments can be tested on
  people only when there is reason to think that
  they will help subjects in the trial
• Future benefits arent enough

27
Clinical trials

• But what about a new treatment that has been
  shown to be completely safe and highly effective
  in animal studies?
• Can we justify not giving it to the half of the
  subjects who will get the placebo in a clinical
  trial?

28
Clinical trials

• Here is what one prominent medical researcher had
  to say about that
• Theres a delicate balance between when to do
  or not do a randomized trial. On the one hand,
  there must be sufficient belief in the agents
  potential to justify exposing half the subjects
  to it. On the other hand, there must be
  sufficient doubt about its efficacy to justify
  withholding it from the other half of the
  subjects who might be assigned to placebos

29
Example 5 Placebo controls?

• We can argue two ways about the following
  question
• Is it ethical to give a placebo to a control
  group if an effective drug already exists?
• The YES side
• A placebo allows us to measure the true baseline
  effectiveness of the drug
• Clinical trials are all different and its
  important to be sure that any new drug is better
  than a placebo, not just better than an existing
  drug that might have been tested against a
  placebo in a previous trial
• Placebo controls are ethical except in
  life-threatening situations

30
Example 5 Placebo controls?

• The NO side
• It isnt ethical to deliberately give patients an
  inferior treatment (when is a treatment truly
  inferior?)
• If there is an effective drug already that has
  already been tested against a placebo, then the
  question is whether the new drug is better, not
  whether it beats a placebo
• Testing against the existing drug whose
  effectiveness already includes the placebo effect
  should be enough
• What if the new drug has fewer side effects for
  many patients but doesnt do as well as the
  existing drug in a trial that just compares the
  two drugs, without a placebo?

31
Example 6 Sham surgery

• This is a very tricky question. Placebos work,
  so why not use them to test surgeries? This
  would involve giving some subjects fake
  surgeries as placebos
• The YES side
• The effectiveness of most surgeries has not been
  tested in comparative experiments, and certainly
  some are just placebos
• Proper placebo-controlled experiments would
  identify these and prevent thousands of
  unnecessary surgeries every year
• The placebo surgeries can be made quite safe

32
Example 6 Sham surgery

• The NO side
• Unlike placebo drugs, placebo surgeries involve
  substantial possible risk to the subject
• Medical ethics require that the the interests of
  the subject must always prevail clearly a
  placebo surgery is not in the interests of the
  subject!
• Even great future benefit does not balance the
  risks to the subject if the subject receives no
  benefit at all
• Sham surgeries would never be used as a regular
  treatment because of their risks, so why should
  they be used as part of a clinical trial?

33
Behavioral and social science experiments
34
Example 7 Keep out of my space

• We all have a personal zone around us that we
  prefer other people do not invade!
• Psychologists want to study this, heres their
  rather annoying experiment
• Set up in a public mens restroom
• Block off several urinals in the middle of the
  row of urinals subjects are forced to use a
  urinal next to an experimenter (the treatment
  group) or at the other end all by itself (the
  control group)

35
Example 7 Keep out of my space

• Another experimenter uses a periscope from a
  toilet stall to observe, the response variables
  are
• How long it takes before a subject begins
  urinating
• How long the subject keeps at it
• This highlights some of the issues involved in
  considering the ethics of behavioral experiments

36
Example 7 Keep out of my space

• No one is physically harmed in this experiment
• In the absence of the possibility of physical
  harm, should we consider protecting subjects
  from
• Emotional distress
• Undignified situations
• Invasion of privacy
• What about informed consent?
• The men in this experiment dont even know they
  are participating in an experiment, but would
  this experiment work if they did?

37
Example 7 Keep out of my space

• Many behavioral experiments would not work if the
  subject knew they were participating in an
  experiment
• The Ethical principles of the American
  Psychological Association require consent unless
  the study merely observes behavior in a natural
  setting
• Deception is only allowed if it is necessary to
  the study, doesnt hide information that might
  affect a subjects willingness to participate,
  and is explained in full to the subjects as soon
  as possible
• The real personal space experiment from the 1970s
  does not meet these ethical standards !!

38
Summary

• Ordinary honesty should prevent people from
  making up data or claiming to be conducting a
  survey when they are really selling something
• Data ethics go beyond simply being honest,
  because even honest people behaving honestly can
  do unethical things
• ALL studies involving human subjects must be
  reviewed in advance by an institutional review
  board (IRB)
• ALL subjects must give their informed consent
  before participating
• ALL information about individual subjects must be
  kept confidential

39
Summary

• These are a good start, but many ethical
  questions remain
• One of the big questions often faced is how to
  balance the welfare of the subject with the
  possible future benefits of the treatment
• A guiding principle that is widely supported is
  the interests of the subject must always prevail
  over the interests of science or society