Letter Regarding the Draft Document Special Ethical Considerations for Epidemiological Research

April 9, 2007

Council for International Organizations of Medical Sciences
c/o World Health Organization
Avenue Appia 20
CH-1211 Geneva 27

To whom it may concern:

Thank you for the opportunity to comment on the current draft of the document Special Ethical Considerations for Epidemiological Research (“the draft document”).   The previous document contained numerous flaws, so a substantial revision is appropriate.

We agree that rationalizing the relationship between the draft document and CIOMS’ 2002 International Ethical Guidelines for Biomedical Research Involving Human Subjects (“the 2002 guidelines”) is appropriate.  However, by in effect making the 2002 guidelines the reference document and providing detail in the draft document only when the 2002 guidelines do not account for special circumstances in epidemiology, the draft document has been made secondary to the 2002 guidelines.  We realize that it is late in the drafting process, but it is more logical and more structurally correct to have the epidemiological guidelines be the reference document because clinical trials are really a special case in the larger category that is epidemiology.

We first identify some elements missing from the draft document and then offer some comments on particular guidelines.

Missing elements

One issue not discussed in the draft document is the possibility of creating a registry for epidemiological studies and their results to parallel those currently being developed for clinical trials.   Requiring researchers to list their datasets and the likely hypotheses they would be exploring would limit both publication bias and inappropriate post-hoc data analyses.

As the draft document makes clear, subject confidentiality is of paramount importance in epidemiological research.   We would like to see more concrete recommendations to prevent situations in which participant identity could be inferred.  For example, the document might recommend that stratification that results in cell sizes of five or smaller generally not be reported if this could render participants identifiable.

Finally, there is a growing industry of epidemiologists who reanalyze data that were generated through government funding.   At least in the United States, such datasets are subject to public disclosure.  Typically, the purpose of such reanalysis is for an industry-funded epidemiologist to cast doubt on a particular finding.  However uncomfortable this may make the government-funded researchers, such disclosure is appropriate and reanalysis does hold the potential for discovering errors or gaining new insights.  But industry-funded studies should be subject to the same degree of disclosure, and the draft document should make that clear.

Guideline 1: Ethical justification and scientific validity of biomedical research involving human beings

The section on research in emergency situations is not adequately prescriptive.   It should make clear that such research should undergo formal prior ethical review except in very unusual circumstances. This change is necessary because, as the guideline acknowledges, “Research that is conducted during an emergency is typically designed prior to the emergency and hence can undergo prior ethical review, at least of its major features” (lines 258-60).

Guideline 2: Ethical review committees

While it may be true that certain research using publicly available or anonymous data may not require review in certain countries (lines 271-3), the present document still has the important role of positing ethical guidelines that may exceed local legal requirements.  As written, the guideline appears to hold the researcher to no standard higher than the local legal one. 

The idea that studies on illicit drug use or domestic violence may merit review (line 279) even if conducted anonymously is inadequate; research on stigmatized populations or that may lead to stigma should automatically be required to undergo ethical review. 

Finally, the issue is not merely if the data are recorded anonymously (line 280), but whether there is true anonymity throughout the research process.  If there are by necessity non-anonymous interactions with study participants from stigmatized groups, there should be a requirement for ethical review even if the data are later recorded anonymously.

Guideline 5: Obtaining informed consent: Essential information for prospective research subjects

It is not sufficient to make reference to the 26-item list in the 2002 guidelines and then say that all of those elements are not required in epidemiological research (line 465).  At a minimum, there should be a discussion of some of the elements more likely to be omitted and of those that must be included, despite a temptation to omit them.  In particular, the seventh item needs to be expanded to discuss in more detail the issue of individual notification to research participants of risks identified in any study in which their data were analyzed.

Guideline 21: Ethical obligation of external sponsors to provide health-care services

This guideline specifies that patients whose conditions are identified in screening studies should receive “pertinent treatment” (line 679).  We would like to see this expanded to clarify that this is a treatment standard consistent with the best science, not with what the prevailing local practice might be, particularly if that practice is limited by poverty.

Guideline 22: Disclosure and review of potential conflicts of interest

The guideline states that researchers should not enter into agreements that “interfere unduly” (line 717) with access to data or the ability to analyze or publish findings.  This is entirely too weak.  The necessary role of a document such is this is to lay down ethical standards with which all research should comply, while providing as little wiggle room for evasion as possible.  At a minimum, this document should specifically preclude, without exception, the following practices:

  • Contracts that are contingent upon reaching particular conclusions
  • Allowing the sponsor to insert its own statistical analyses into articles
  • Drafting of the manuscript by anyone but an author
  • Unreasonable limits on the ability of investigators to publish
  • Unreasonable limits on the ability of investigators to share data with other investigators

Despite the guideline’s claim to the contrary, if an ethical review committee is paid to review a study, this does represent “an inherent conflict of interest” (line 732-3).  Because for-profit ethics committees need to continue to attract business, an outcome not fostered by rejections of research protocols, such ethics committees should be utilized only when there is no satisfactory alternative, such as a university or government committee.  The guideline should explicitly criticize the practice of “ethics committee-shopping.”

Guideline 24: Use of stored biological samples and informed consent

The final element of this guideline (lines 822-6) says that stored samples collected without informed consent could be used if the rights or welfare of the donors were not harmed.  An additional condition should be that the research be clearly documented to be in the public interest.

Thank you again for the opportunity to comment on this draft.

Yours sincerely,

Peter Lurie, M.D., M.P.H.
Deputy Director 
Public Citizen’s Health Research Group

Sidney M. Wolfe, M.D.
Public Citizen’s Health Research Group