|
Epi Wit & Wisdom Articles
Ethics in Epidemiology
Ethics is Focus of Two Day
Meeting in Birmingham
A conference on ethics in
epidemiology was held in June in Birmingham immediately preceding the
annual Society for Epidemiologic Research meeting. Sponsored by the
Industrial Epidemiology Forum (IEF), the meeting was the largest of
its kind ever held for epidemiologists. Over a two day period, almost
200 epidemiologists from various fields listened to over two dozen
presentations on different aspects of ethical issues as they pertain
to epidemiology. According to Dupont epidemiologist Bill Fayerweather,
organizer of the conference, the purpose of the meeting was to raise
awareness about ethical issues in epidemiology, and the conference was
noteworthy for the number of issues raised and presented, ranging from
traditional concerns about conflict of interest to the less seldom
heard concerns about biases in publishing epidemiological findings.
Because of the uniqueness of
this meeting and the importance of this topic for epidemiologists in
all subspecialties, this issue of the Epi Monitor is devoted to
recapping highlights from the presentations. Full publication of the
proceedings of the meeting is expected at a later date.
Bioethics and Epidemiology
Chaired by Susan Sacks from
Syntex, the first session provided an overview of the philosophical
framework underlying many of the presentations that would follow. Tom
Beauchamp, an ethicist from Georgetown University, presented the first
talk on ethical theory and described what it can and cannot contribute
to a discussion of ethics and the development of a code of ethics for
epidemiologists. For example, he stated that ethical theory can
provide general principles which can then lead to more specific
guidelines, such as the principle of respect for autonomy leading to
guidelines on informed consent. However, he cautioned that ethical
theory “gives arguments not answers.” He likened moral reasoning to
legal reasoning and said that arguments can be made well or poorly,
and that counter arguments can always be made, and are, in fact,
expected.
In the same session, Hopkins
epidemiologist Leon Gordis expressed his view that epidemiology is not
an isolated intellectual exercise and cannot be defined in terms of
only one activity. Rather, a constellation of activities constitute
epidemiologic practice. In his view, epidemiology reaches beyond
research and has an inherent relationship with policy development.
This is because epidemiologic findings are of direct relevance to
society, are paid for with public funds, involve human subjects, and
are not always of benefit to the research subjects themselves. For all
these reasons, epidemiology may differ from other research pursuits
and unique ethical issues are raised.
NCI’s Douglas Weed also spoke in
this session about the merger of bioethics and epidemiology. He stated
his belief that the time has come to pursue a philosophy of
epidemiology and that the merger of bioethics and epidemiology could
be the cornerstone of this development.
Epi Research in Different
Contexts
In this session, University of
Pennsylvania epidemiologist Paul Stolley presented the opinions of an
SER committee on conflict of interest issues, Epidemiology editor
Kenneth Rothman took issue with a NEJM policy requiring investigators
to identify their sponsors when publishing results, Dow epidemiologist
Gregory Bond discussed an epidemiologist’s obligations to subjects,
society and employers, and University of Alabama epidemiologist Philip
Cole discussed the pros and cons (mostly pros) of serving as a paid
expert witness in court.
Among the conflict of interest
guidelines proposed by Dr. Stolley’s committee were 1) the
independence of the investigator(s) must be maintained; 2)
epidemiologic research should not be conducted secretly and results
should receive timely publication; 3) the release and publication of
data should not be affected by whether or not the sponsor views the
results as favorable or unfavorable; 4) all sponsorship of research
should be publicly acknowledged and there is no justification for
secrecy in regard to sponsorship; 5) all investigations, whether
government or industry-sponsored, are the intellectual property of the
investigators, not the sponsor. The decision as to how or when to
publish or otherwise disseminate data should be the sole
responsibility of the investigator.
Of particular concern to Dr.
Rothman are the implications of asking investigators to identify their
sponsors. He believes this attacks the integrity or competence of the
investigators. “The identity of one’s sponsor is irrelevant and what
is needed are impartial evaluations of research findings based only on
the methods and data alone,” said Rothman. To the extent that
identification of sponsors focuses the evaluation of research on
potential conflicts of interest, this detracts from objective reviews,
according to Rothman.
In the same session, Dr. Bond
highlighted how all epidemiologists have potentially conflicting
obligations to different parties. In his view, the goal should be to
recognize this in advance, and to design research programs and
procedures which avoid or keep conflicts to a minimum. He cited
examples from Dow Chemical where peer review policies, development of
an IRB within the company, and prior commitments to publish regardless
of findings are used to minimize conflicts of interest.
The final speaker to address
epidemiologic issues in different contexts was Dr. Cole who made a
provocative presentation about the challenges of serving as an expert
witness. Contrary to the popular wisdom that serving in this capacity
is unpleasant and demeaning, Dr. Cole described the court as a “place
to practice with pride.” He depicted the court as a place where
“epidemiology comes alive,” where one has the opportunity “not only to
pursue truth but to advocate it.”
He said that epidemiologists
should “not fear the courtroom any more than the classroom, and...you
have a vital role to play in society which is uniquely yours to play.”
He stated his view that money
does not cause epidemiologists to perjure themselves or to express
biased opinions, but rather that holding a particular opinion in
advance is what causes epidemiologists to get paid to articulate these
views in court. He suggested that epidemiologists “look down” on paid
witnesses because they have not grasped the true, genuinely positive
motives for testifying. “Ignore the implication (from being a paid
witness)--so will everyone else,” he said.
Communicating Health Risk
Information
The third session on the
afternoon of the first day included the largest number of speakers. A
colorful talk was given by Peter Sandman, a Rutgers communications
expert, on the topic of communicating risk information. He described
his work as alerting people to risks they are likely to ignore--and as
reassuring people about risks that are overestimated. In simpler
terms, scaring people and calming people down--but not at the same
time!
Dr. Sandman presented several
principles to keep in mind in communicating risk information from
epidemiologic studies. They are:
1) Tell the affected group in a
timely fashion what you have found.
This responsibility is not met
by putting a report in a professional journal, according to Dr.
Sandman. People do not wish to be told about their risks in the mass
media. He joked, “You don’t want to hear you’re dying on the six
o’clock news!”
2) Make sure people understand
your findings and their implications.
To accomplish this, he suggested
simplifying the content of findings and cutting out jargon. All
scientists hate this, he said. But either you will simplify or it will
be done by the media or the public. The choice is clear.
Also, to make people understand,
epidemiologists need to interpret their findings for others. This is
difficult because it takes epidemiologists outside of epidemiology,
and many consider it none of their business. But Dr. Sandman urged
epidemiologists to go beyond their field and to address the most
obvious and appropriate questions which arise from the point of view
of the public.
3) Find ways to bolster
credibility.
Dr. Sandman urged
epidemiologists to do things in advance of releasing results such as
insulating themselves from their sponsor by the use of review boards,
and obtaining the participation of the community in the study. The
results will be easier to believe if the community participates in the
design and implementation of the study, he said.
4) Acknowledge uncertainty.
If investigators do not
acknowledge uncertainty, it makes them appear incredible. In any case,
your peers will point out uncertainty if you do not, and you are more
believable if you are the first to point this out.
5) Apply epidemiology where it
is called for and do not misapply it where it will not help.
For example, avoid studies with
low power. Avoid studies which cannot answer questions.
6) Show respect for the public.
In this regard, Dr. Sandman
reminded the audience that people trust anecdotal data even if
epidemiologists do not. “People do not like to hear that Aunt
Matilda’s cancer was an outlier!” he said.
He also encouraged
epidemiologists to draw the distinction between statistical and social
significance, and finally to respect and acknowledge the public’s
genuine outrage when there has been lack of fairness or truthfulness.
In another discussion of
communicating health risk information, Georgetown’s John Higginson
made the similar point that communication will not succeed if the
public is not conditioned to accept it.
Paul Schulte from NIOSH
described some of the ethical issues involved in notifying workers
about study results. Special problems are posed by uncertain test
results and by the financing of medical follow-up. What level of
resources should be used to find and notify workers? Will the
obligation to notify workers cause some investigators to lose interest
in occupational studies? NIOSH has been providing some guidance on
these issues.
Gary Spivey from Unical gave his
views about how tenets of ethical theory can provide some guidance in
communicating health risk information. Respect for autonomy should
lead epidemiologists to protect confidentiality and to disclose
information to subjects so that they can make their own decisions. The
principle of beneficence leads one to communicate health risks to the
persons affected, and to work to build trust with the subjects. The
ethical principle of non-maleficence dictates that one not
under-interpret or over-interpret risks.
Albert Jonsen, a professor of
ethics in medicine at the University of Washington, presented the view
that communicating health risk information in terms of everyone’s
“rights” may be unnecessarily immobilizing and confrontational. He
proposed instead the view that information about one’s environment is
a vital constituent of modern life, just as vital as oxygen. Those who
lack it or cannot use it are disabled. In this view, hazard
information represents the interpreted experiences of everyone and is
the common property of all. Proprietary rights do not apply here, and
the rule should be that special justification is needed to conceal or
withhold information, not vice-versa.
Responsibilities to Research
Subjects
Among the issues discussed by
Manning Feinleib from NCHS were the appropriateness of using financial
incentives in studies where informed consent is obtained, the amount
of effort which is justified in seeking to change the minds of persons
who have refused to participate in a study, and the analytic
safeguards which should be followed to prevent inadvertent disclosure
of patient identities.
Alexander Capron from USC urged
epidemiologists not to dismiss ethics as just an obstacle to their
work. He noted that harm from epidemiological studies is rarely
physical and may not take the same form as in other biomedical
pursuits. He cited examples of harm coming from workers being
ostracized with subsequent loss of employment and harm coming from
populations being painted negatively.
William Thar from Exxon noted
the close relationship which epidemiologists have with their employer
in the occupational studies setting. In this context, obtaining
informed consent should be viewed as the beginning of a communication
process between the epidemiologist and employees. In this setting,
workers should be given a clear timetable for the study, should be
told about sponsorship and about the qualifications of the
investigator, plans for disseminating the information should be
described, and the reasons for doing the study explained. According to
Thar, “informed consent in the occupational setting should be an
ongoing, dynamic process with the research participants.”
John Last from the University of
Ottawa noted how respect for the autonomy of individuals is considered
one of the most basic values in our society and that obtaining
informed consent is how this principle is applied in practice in
research settings.
Data Access, Interpretation,
Presentation, and Publication
Carol Hogue from CDC served as a
member of an SER committee looking into data sharing and presented the
opinions of this committee. Multiple reasons can be advanced both in
favor of or in opposition to sharing data. The SER committee has
concluded that the decision to share data should not depend on the
source of the data (i.e. public or private), that release should not
depend on the presumed motivation of the requester, and that it is the
primary responsibility of the recipient to carry the burden of sharing
data (e.g., costs of replicating documents). Some of the issues
remaining less clear are what should be the timing of the release
relative to original publication, and should raw data with identifiers
be released. The group concluded that institutions should develop
policies on sharing data and that there should be an appeals process
for requests that are denied. In conclusion, the group felt that data
sharing is a complex issue and much further discussion will be needed.
In the meantime, the idea of data sharing is endorsed if the rights of
the investigators can also be protected.
Moyses Szklo, editor of the AJE,
discussed the biases which can operate in the dissemination of
epidemiologic findings both before and after the results are
published. For example, pre-publication bias exists when factors other
than the manuscript quality dictate the acceptability of the findings
for publication. Examples of this would include the systematic
tendency to publish positive as opposed to negative findings. Dr.
Szklo discussed the creation of study registries as one way to prevent
this problem.
Geoffrey Paddle from Imperial
Chemical Industries reflected on the situation in the UK where he said
ethics is not a big issue. One area of potential concern is the
investigation of multiple hypotheses different from the original
hypotheses. He cited the example of studies which look at standardized
mortality ratios. If not elevated, we try other subgroups, he said.
This is the “stop at the winner approach.” He noted that
interpretation of findings is theoretically predictable from the
objectives of a study and that this point is important in industry
which does not like surprises. In his view, multiple objectives create
new responsibilities, so that studies with only one or two objectives
are easier to deal with.
In comments on Dr. Paddle’s
talk, Dr. Jonsen stated that at the heart of efforts to develop a code
of ethics for epidemiologists is the need to determine what are our
allegiances. Do these allegiances have priorities? To the truth? To
the social welfare? To the employers? What is epidemiology all about?
When there is communal agreement about these issues, epidemiologists
can develop a code of ethics, according to Jonsen.
Approaches to the Future
In the sixth session,
chairperson Sandra Tirey from the Chemical Manufacturers Association
gave her opinion that “ethics in epidemiology ultimately boils down to
a question of credibility--is epidemiology a reliable means of
answering questions about human health?”
Alvan Feinstein from Yale urged
epidemiologists to keep conflicts in science separate from conflicts
about ethics. He noted that historically scientific controversy often
leads to “passion,” i.e., when a scientific argument is threatened,
ethical arguments arise.
He made these remarks by way of
preface to the real point of his talk which was that the basic
paradigms of epidemiology need revision, that these changes will cause
controversy, and that accusations of unethical behavior are not
helpful. He bemoaned the lack of a set of standards for carrying out
studies and called some current practices “shoddy.” He expressed
confidence that if the changes he perceives as necessary are
worthwhile, they will succeed.
Colin Soskolne from the
University of Alberta, who was among the first in the epidemiology
community to raise ethical issues publicly a few years ago and to call
for a code of ethics, used a case study approach in his talk to
illustrate how ethical decision making is a process that requires
knowledge of the specific situation as well as knowledge of ethical
principles. “It’s a process that challenges at every step,” he said.
John Andrews from the Agency for
Toxic Substances and Disease Registry discussed the peer review
process in his presentation and concluded that peer review as
currently practiced has problems. It is conducted only after the study
is completed, and is done by busy researchers who cannot tell if the
research was well done. He proposed carrying out peer review at the
conception of a study, while it is ongoing, and at the completion of a
study. Good peer review could help to insure good epidemiology, he
stated, and persons participating should be reimbursed because of its
potential benefits.
In his discussion of approaches
for the future, Dr. Ralph Cook from Dow Chemical presented a draft
code of ethics which was prepared by a subcommittee of the Industrial
Epidemiology Forum. Dr. Cook noted that the pressures on epidemiology
have changed over time. In the past, researchers and users tended to
be the same, whereas now the users of epidemiologic data are not only
the researchers, but persons outside of epidemiology as well. As the
scope of epidemiology broadens, “we must augment the checks and
balances,” he said. He sees a code of ethics as one part of the
solution, in addition to audits and awards, which will serve to
promote sound epidemiology in a positive rather than negative way. He
views the code proposed by the IEF as an early draft and believes the
profession is in the early stages of a complex dialogue.
Recapitulation
In this final session of the
meeting, Irving Selikoff from Mt. Sinai and Brian MacMahon from
Harvard provided a recap of the meeting. Dr. Selikoff provided a
historical recounting of the development and evolution of the concept
of prevention.
Dr. MacMahon noted in closing
that the range of topics discussed under the banner of ethics is very
broad, and stated his view that ethics is the wrong word for what
concerns epidemiologists at the present time. He believes that ethical
conduct rather than ethics per se is what is on the minds of
epidemiologists.
Dr. MacMahon expressed the view
that progress in this area would come one step at a time. There has
been an unwritten code of ethical behavior up to now and the question
being faced is whether or not it is feasible and desirable to capture
this code in written form. He noted that he is inclined to agree with
the need for a written code, however, he stated that it should be a
short, flexible and practical code dealing with real life issues that
are practical problems. Enforcement is not essential in his view, but
broad participation in the development of the code is very desirable.
Published June 1989 v
|
