American Journal of Law & Medicine

The Ethics of Postmarketing Observational Studies of Drug Safety under Section 505(o)(3) of the Food, Drug, and Cosmetic Act

     A. Novel Aspects of Section 505(o)(3)
     B. FDAAA's Hierarchy of Approaches for Studying
        Postmarketing Drug Safety Questions
     A. Access Under the HIPAA Exception for Disclosures to
        Drug Manufacturers
     B. Access Under the HIPAA Exception for Disclosures to
        Public Health Authorities
     C. Access Under the HIPAA Waiver Provision
     D. Impact of State Privacy Laws
     A. The Section 505(k)(4) Data Access Pathway
     B. The Ethics of Data Access Under Section 505(k)(4)

In 2007, Congress granted the Food and Drug Administration (FDA) new powers to order pharmaceutical companies to conduct drug safety studies and clinical trials in the post-marketing period after drugs are approved The methodologies include observational studies that examine patients' insurance claims data and clinical records to infer whether drugs are safe in actual clinical practice. Such studies offer a valuable tool for improving drug safety, but they raise ethical and privacy concerns because they would entail widespread use of patients' health information in commercial research by drug manufacturers. This is the first article to explore the ethics of these section 505(o)(3) observational studies, so named after the section of the Food, Drug, and Cosmetic Act that authorizes them.

Data access problems threaten to make the FDA's section 505(o)(3) study requirements unenforceable. Under existing federal privacy regulations, it appears highly unlikely that pharmaceutical companies will have reliable access to crucial data resources, such as insurance claims data and healthcare records, to use in these studies. State privacy laws present another potential barrier to data access. If pharmaceutical companies do manage to gain access to the needed data, this will raise serious privacy concerns because section 505(o)(3) observational studies do not appear to be covered by any of the major federal regulations that afford ethical and privacy protections to persons whose data are used in research. If the FDA's program of section 505(o)(3) observational studies fails because of the above problems, this failure will have a number of bad consequences: the public will be exposed to avoidable drug safety risks; taxpayers may be forced to bear the costs of having the FDA conduct drug safety investigations that would have been funded by drug manufacturers if data had been available; and, perhaps most troubling, the FDA may be forced to order postmarketing clinical trials to answer questions that could have been answered using observational studies. Problems with access to data for section 505(o)(3) studies thus could directly imperil human research subjects by forcing a needless over-reliance on risky postmarketing drug safety trials. This Article concludes by describing a promising new legal pathway for resolving these problems. Congress has provided the FDA a new set of powers that if skillfully exercised will allow the agency: (1) to facilitate pharmaceutical companies' appropriate access to data for use in section 505(0)(3) observational studies, (2) to impose strict ethical and privacy protections for persons whose data are used in these studies, and (3) to mobilize private-sector funding to generate much-needed evidence of the safety of FDA-approved drugs.


In April 2011, the U.S. Food and Drug Administration (FDA) published its final guidance on Postmarketing Studies and Clinical Trials ("Guidance"). (1) The Guidance clarifies how the agency plans to exercise its new powers to require pharmaceutical companies to investigate safety issues with already-approved prescription drugs and biologics (together, "drugs"). The Food and Drug Administration Amendments Act of 2007 (FDAAA) (2) created these powers by adding a new section 505(o)(3) (3) to the Federal Food, Drug, and Cosmetic Act (FFDCA). (4) In the postmarket period after a drug is approved, section 505(o)(3) allows the FDA to order manufacturers to conduct various types of studies including animal and laboratory studies, clinical trials, (5) or observational (6) studies. (7)

These provisions give the agency an important new tool to improve clinical drug safety and protect the public's health, but they raise potential ethical and privacy concerns. There are many kinds of observational studies. They vary both in their privacy implications (How much and what types of data do they require? (8)) and in their evidentiary value (How reliably do they answer the scientific question at hand? (9)). Some observational studies are ethically uncomplicated. An example would be a prospective study that records clinical observations to follow how patients are responding to a drug that they chose to take as part of their clinical care. (10) It is a simple matter to obtain consent of the participating individuals while researchers are meeting with them to observe their medical progress.

Other types of observational studies raise thornier ethical and privacy concerns. Modern cohort (11) and registry (12) studies often rely on large administrative databases (such as claims data held by insurers) and clinical data (such as medical records held by large hospitals or health maintenance organizations). (13) Examples include large scale retrospective studies that "repurpose" historical insurance claims data and clinical records from patients' actual encounters with the healthcare system to infer whether drugs are as safe in clinical practice as the FDA expected them to be when it approved them. (14) The Guidance envisions ordering these types of section 505(o)(3) observational studies (15) To be able to perform such studies, pharmaceutical companies would need broad access to patients' health data. In large-scale studies of this type, obtaining consent may be impracticable or may bias the dataset in ways that would reduce the scientific validity of the findings.16 Such studies therefore may entail nonconsensual use of people's health data.

The decision to order such studies implicates individual rights and competing values-such as drug safety and public health-that could be advanced by allowing access to people's sensitive health data. This itself is nothing new: privacy sometimes does collide with other important public interests. When it does, there is wide agreement among bioethicists that the "central ethical issue" is to ensure that the potential public benefits are sufficient to warrant the burden on the individual. 17 What is new-and what this Article explores-is that section 505(o)(3) presents an additional problem of comparative research ethics.

Section 505(o)(3) recognizes that, in the postmarket environment, there often are multiple methodologies available for answering a given drug safety question. A randomized, controlled clinical trial (RCT) is one way to study such questions. Large-scale observational studies are another way. In the postmarket setting, the FDA faces choices among multiple available methodologies. In contrast, the FDA has a narrower set of methodologies available for premarket studies. The drug is not yet on the market; there are no people already taking it; thus, there is no population of drug consumers to "observe." The only option is to conduct a clinical study that deliberately introduces the drug into a group of test subjects. The postmarket setting offers a richer field of available methodologies, including the possibility of drawing inferences from the experience of consumers already taking the drug.

Large-scale observational studies, if carefully designed, can generate rather high-quality scientific evidence, (18) RCTs are generally viewed as the most reliable source of medical evidence. Concato et al. argue that the perceived superiority of RCTs arose, in large part, from a landmark 1982 study (19) comparing RCTs to observational studies, (20) The 1982 study had considered a group of observational studies that used historical controls (in other words, the observational studies had compared data for current patients to data for patients in the past). That fact alone may have contributed to the poor perceived performance of the observational studies: historical differences in treatment methods may have confounded their results. 21 Well-designed observational studies that use contemporaneous control subjects can provide results that are "remarkably similar" to RCTs. (22) RCTs do have undeniable evidentiary advantages in certain contexts-particularly when studying drug efficacy (23) -but observational studies offer fairly high-quality evidence without exposing human subjects to the risks of clinical trials, and they may even be superior to RCTs when studying certain types of safety issues. (24)

There are thus two central ethical issues in postmarketing drug safety studies under section 505(o)(3). The first is the traditional question of whether to do a study at all: Is the knowledge to be gained worth the burden on participating individuals? The second question is which methodology to use: an observational study that potentially inflicts privacy or dignitary harms or a clinical trial that potentially may place human subjects at real physical risk. The alternatives expose individuals to different risks and ethical burdens and they have different evidentiary advantages and disadvantages. Even when knowledge is clearly worth gaining, that is not the end of the ethical analysis, which also must ask, What is the least burdensome way to gain the knowledge, and what ethical price are we willing to pay for incremental knowledge? Protecting privacy by refusing access to data for a section 505(o)(3) observational study may place human subjects at risk in clinical trials that could have been avoided if the observational study had been possible.

It is overly simplistic to portray the ethics of postmarketing studies as a binary choice in which denying access to data protects ethical values at the expense of practical objectives such as improving drug safety or public health. The dilemma is more complex. Denying access to data may promote ethical values at the expense of ethical values: well-motivated efforts to protect privacy may have the effect of exposing human subjects to avoidable research-related risks.


Since 1962, the FDA has been able to require drug sponsors to conduct premarket clinical studies, including RCTs, to show that drugs are safe and effective and thus worthy of an FDA approval. (25) FDAAA gave the agency significant new powers that have been described at length elsewhere. (26) To summarize, the FDA's powers to require premarket studies remain unchanged, (27) but Title IX of FDAAA greatly expands the agency's regulatory powers during the postmarketing phase of the drug lifecycle. FDAAA pragmatically recognizes that premarket drug trials, even when well designed, will never be able to detect all of a new drug's potential safety issues because of problems with trial size, trial duration, and non-generalizability. (28) FDAAA expands the FDA's authority to require drug sponsors to study drugs after they are approved. (29) These new powers under section 505(o)(3) are, however, only one component of FDAAA's overall program to enhance drug safety. FDAAA also expands the FDA's own capacity to conduct postmarketing surveillance and observational studies of drug safety by authorizing creation of a nationally scaled health data network (30) to support such activities. This postmarketing "risk identification and analysis system," (31) commonly known as the Sentinel System, (32) is a large, distributed health information network that hamesses administrative data and clinical records. (33) The Sentinel System's pilot project, Mini-Sentinel, already includes data for 99 million persons (34) and is returning results in response to drug safety queries from the FDA. (35)

FDAAA does not treat evidence generation as an end in itself and expands the FDA's power to take concrete regulatory actions in response to new evidence of safety problems The agency can require safety-related labeling changes, (36) although FDAAA also recognizes that traditional labeling is an imperfect way to communicate and authorizes an Internet-based system (37) for sharing risk information (38) with patients and providers. FDAAA also authorizes the use of Risk Evaluation and Mitigation Strategies (REMS), (39) which are plans for evaluating and managing a drug's safety risks. The agency can require a REMS either when a drug is first approved (40) or later if the agency finds, based on new evidence, that a REMS is necessary to ensure that the drug's benefits outweigh its risks. (41) If the FDA requires a drug to have a REMS and the drug sponsor fails to comply with its terms, sale of the drug becomes unlawful. (42) All REMS must include plans for ongoing evaluation of the drug's risks. (43) A REMS also may include additional elements. (44) In the most severe circumstances, these can include "elements to ensure safe use" (45)--actual restrictions on the sale, distribution, or use of a drug that is found to have such serious risks that it otherwise would need to be kept off the market. (46)

The expansion of premarket clinical drug trial activity after 1962 was a factor that fueled pressure in the 1970s to strengthen federal protections for human research subjects. (47) FDAAA anticipates a major expansion of postmarket evidence generation, both by the FDA and by drug manufacturers. Because FDAAA calls for heavy reliance on observational studies as a tool for studying drug safety questions, it implies a similar need to strengthen and rationalize protections for people whose data will be used.

There is a fairly substantial body of literature addressing ethical and privacy issues in the FDA's own observational studies using the Sentinel System. (48) The scholarly community has been slow to engage with the ethical and privacy aspects of section 505(o)(3) observational studies conducted by drug manufacturers. (49) The remainder of this Article surveys the aspects of section 505(o)(3) that create privacy concerns and other provisions of FDAAA that offer a promising legal pathway for addressing those concerns. The goal of this discussion is to stimulate wider debate about how best to fulfill the important public health objectives of section 505(o)(3) without unduly burdening the people whose privacy may be affected.


Section 505(o)(3) requires the FDA to choose which scientific methodology is best for answering specific drug safety questions. (50) The FDA will have to decide when to rely on an observational study and when to insist on an RCT. (51) The agency also will have to choose among various statutory mechanisms for arranging an observational study. (52) The statute offers various alternatives and constrains the agency's choices among them. (53)


Section 505(o)(3) confronts bioethicists with a strange new hybrid: commercial research that serves a public health objective. On one hand, section 505(o)(3) investigations are commercial research, since drug manufacturers will perform or at least fund them. On the other hand, a governmental public health agency (the FDA) will order these investigations to address public health concerns about the safety of approved drugs. Section 505(o)(3) allows the FDA to order studies and clinical trials at any point after a drug is approved, but only in response to new safety information that has emerged while the drug was in clinical use (54) and only for narrowly circumscribed purposes, (55) The allowed purposes are to assess known serious risks related to the use of a drug, to assess signals that suggest a drug may have serious risks, or to identify unexpected serious risks when the available evidence indicates the potential for a drug to have a serious risk. (56) Although public health concerns motivate the FDA to require these investigations, the sponsors will be drug companies acting with clear commercial motives. Failing to perform a required section 505(o)(3) investigation can make it unlawful to ship the drug in interstate commerce, can cause the drug to become "misbranded" so that selling the drug is prohibited, and may result in large civil monetary penalties, (57)

Another noteworthy aspect of section 505(o)(3) is its terminology, which distinguishes postmarketing studies from postmarketing clinical trials. …

Log in to your account to read this article – and millions more.