Customize your JAMA Network experience by selecting one or more topics from the list below.
In this issue of JAMA Internal Medicine, Kernan et al1 chronicle Yale University’s experience responding to a subpoena for data from an ongoing, double-blind, placebo-controlled trial of pioglitazone. The subpoena arose from litigation brought by Sara J. Kincaid,2 who believed she had been injured by pioglitazone but who was not a clinical trial participant. Yale’s legal team was troubled because they believed that releasing the data would compromise the integrity of the trial and threaten the investigators’ scientific interests.
The authors’ account of their struggle to balance their legal obligations with fidelity to good science is fascinating. Herein we provide broader context to this important issue by examining the subpoena power and the authors’ conclusion that there is a “duty to resist” subpoenas for data from ongoing clinical trials. Two considerations qualify such a duty. First, intermediate solutions may exist between publicly disclosing and refusing to turn over clinical trial data. Second, the balance of the various interests involved in weighing responses to a subpoena tips in different directions depending on the stage of the clinical trial.
The reaction of people who are not lawyers to Yale’s predicament may be to ask “Why can third parties be burdened with subpoenas when they have nothing to do with a plaintiff’s grievance?” The legal system in the United States embodies a strong preference that cases be decided correctly, on the basis of all available facts, although pursuing this goal may heavily burden the parties to the case and others. Providing a means to compel production of evidence is necessary; plaintiffs bear the burden of proof, yet defendants and others who have no incentive to share information typically have a considerable amount of the information that plaintiffs’ attorneys need to build their case. Discovery is the name given to the process of seeking evidence and requiring its production, and the mechanism for compelling those who are not parties to the litigation to disclose documents is called a subpoena duces tecum.
The scope of what may be discovered is extremely broad. For example, in federal court, according to the Federal Rules of Civil Procedure (Fed R Civ P), litigants may obtain information about virtually any matter relevant to their claim or defense, so long as what they seek is “reasonably calculated to lead to the discovery of admissible evidence” (Fed R Civ P 26[b]). However, some safeguards prevent plaintiffs’ lawyers from abusing discovery to embark on “fishing expeditions” without reasonable basis. There are some limits on what may be discovered (only things relevant to claims and defenses, and only materials not shielded from discovery by a legal privilege, such as attorney-client privilege). Following established standards, judges can issue protective orders limiting the extent of discovery or restricting how evidence is disclosed (Fed R Civ P 26[c] and [d]). When subpoenaed parties ask courts for relief, judges weigh the burdens of producing the information against the plaintiff’s need for the information and its probative value.3
In the Kincaid case,4 there appears to have been no dispute that the subpoenaed information was relevant to the plaintiff’s claims. Yale’s objections pertained to the effect that disclosure would have on the ongoing trial, the investigators, and the public. How compelling were Yale’s objections?
We agree with Kernan et al1—and the judge who ruled in Yale’s favor4—that a framework of balancing risks and benefits is appropriate for determining lawyers’ obligations in responding to subpoenas for clinical trial data. We also agree that harm may follow from unhesitating compliance with subpoenas. However, the risk of particular harms varies depending on the stage of the clinical trial at the time the disclosure is requested: (1) during data collection, (2) between study closure and the first publication, or (3) after the first publication (Table). The risk of harm to research participants, investigators, and the public is highest when data collection is ongoing, the first stage. Particularly grave is the risk of undermining the integrity of the trial, causing a loss to science. The balance of interests militates in favor of declining to release data or requiring the plaintiff to narrowly tailor the discovery request to capture the minimum necessary amount of information and show why the request cannot wait until data collection is complete.
Judges should also consider whether procedural safeguards could sufficiently minimize the risks associated with the disclosure of data. For example, as was case in the Prempro litigation,5 a protective order could (1) prohibit publication and discussion of the study data or analysis outside of court; (2) specify who may access the data; (3) limit the data fields disclosed; (4) seal court submissions; (5) delay the release of treatment assignment information; and (6) require plaintiffs and their experts to adhere to these conditions under penalty of civil sanctions. Protective orders do not eliminate the risk of public dissemination, since they may be violated,1,3 but they substantially shift the risk-benefit balance.
More creatively, judges could appoint a “special master” or other neutral third party to analyze the unblinded data without releasing them to the plaintiff. This could also address a major concern raised by Kernan et al1—that complying with the subpoena would have meant unblinding the Yale investigators. A skilled intermediary can link data sets containing participants’ clinical information and treatment assignment so that the investigators themselves do not become aware of information about treatment assignment.
Between the closure of data collection and the first publication, the second stage, the primary risk is that the plaintiff’s attorney will mischaracterize trial results in public communications. Typically, the first publication reports the main results of a study. A secondary risk is the incursion on the trial investigators’ right to be the first to report these results. This risk is real because, and contrary to the characterization by Kernan et al,1 the data are not “public goods.” Public goods are nonrivalrous—meaning that their use by one person does not diminish their utility to others—but that is not the case for research data before the first publication.
In the second stage, the balance of interests weighs more in favor of releasing data than it does during the collection of data and might support a broader disclosure. What is critical is precluding public dissemination, not preventing disclosure to the plaintiff. Sealing the disclosed documents, therefore, may be preferable to resisting the subpoena. Judges could also balance the plaintiff’s and investigators’ needs by giving investigators a specified period in which to publish results, granting the plaintiff’s discovery request thereafter, and holding the case in abeyance in the interim.
The third stage begins with the investigators’ initial publication of results. At this stage, subpoenas can be thought of as just another way of accomplishing what we and others have argued should generally occur for completed trials: broad access to deidentified data.6 Such data should be accessible to any qualified user with a legitimate reason for access. Potential harms associated with the release of data are manageable, and expanded access to data can accelerate the progress of science.7
Clinical investigators and the sponsors of clinical trials may worry, with reason, about protecting the privacy of research participants and how attorneys may use and characterize the data. Thus, the focus after the first publication should be on ensuring responsible use of the data and preventing the reidentification of research participants.7 Courts could, for example, require plaintiffs’ attorneys to justify a need for each piece of information requested that could potentially be used to identify the research participants, and to agree not to attempt reidentification. Judges should also apply existing discovery rules to ensure that data requests are reasonably well founded and not mere fishing expeditions.
In the Kincaid case,4 Yale and its investigators had a duty to protect the integrity of their research. The justice system in the United States has a duty to provide a forum where truth will emerge and wrongs can be remedied. These duties can best be reconciled by acknowledging that the balance of interests shifts over the life cycle of a clinical trial.
Corresponding Author: Michelle M. Mello, JD, PhD, MPhil, Stanford Law School, 559 Nathan Abbott Way, Stanford, CA 94305 (firstname.lastname@example.org).
Published Online: July 21, 2014. doi:10.1001/jamainternmed.2014.2341.
Conflict of Interest Disclosures: Dr Mello serves as a consultant to CVS/Caremark on the issue of tobacco-free pharmacies.
Funding/Support: Dr Mello’s work was supported by a Lab Fellowship from the Edmond J. Safra Center for Ethics at Harvard University, and Mr Cohen’s work by a Greenwall Faculty Scholars Award.
Role of the Sponsor: The sponsors had no role in conception or creation of this commentary; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Information: Dr Mello is now with Stanford Law School and Stanford Medical School, Stanford, California.
Mello MM, Cohen IG. Clinical Trials and the Right to Remain Silent. JAMA Intern Med. 2014;174(9):1505–1506. doi:10.1001/jamainternmed.2014.2341
Create a personal account or sign in to: