Post Market drug Safety: the view from FDA
Laurie E. Scudder, DNP, NP
May 19, 2017
Two recent studies have raised questions about postmarketing safety issues with newly approved drugs, particularly those granted approval on the basis of limited data. The first study, conducted by Alison M. Pease, a medical student at State University of New York Downstate College of Medicine in Brooklyn, and published in the BMJ, used the Drugs@FDA Database to identify all novel drugs that were approved by the US Food and Drug Administration (FDA) between 2005 and 2012 on the basis of either a single pivotal trial or pivotal trials focused on surrogate markers of disease. During this period, more than a third of new approvals were awarded on the basis of a single pivotal trial; just under half (44%) were approved on the basis of studies that used surrogate disease markers rather than clinical outcomes. The researchers then conducted a systematic review of prospective controlled clinical studies published after approval. For 35.0% of these newly approved indications, the researchers were unable to identify any postmarket clinical trials. A median of one to three postapproval studies was identified for agents or indications approved on the basis of a single pivotal trial, and the aggregate number of patients enrolled in these postapproval studies was 90.
The second study, conducted by Nicholas S. Downing, MD, from the Department of Medicine at Brigham and Women’s Hospital in Boston, Massachusetts, and colleagues, and published in JAMA in May 2017, also used data from the Drugs@FDA database to identify postmarket safety events reported for 222 novel therapeutics approved by the FDA between 2001 and 2010. Over a median follow-up period of 11.7 years, postmarket safety events were reported for 32% of these agents. These safety issues led to three drug withdrawals.
To discuss the issue of postapproval monitoring of newly approved agents, Medscape spoke with Gerald Dal Pan, MD, MHS, director of the Office of Surveillance and Epidemiology at the FDA’s Center for Drug Evaluation and Research, about the postmarket approval process and its implications for clinicians prescribing newly approved agents.
Evaluating Safety Through the Lifecycle
Medscape: Can you describe the lifecycle evaluation process and the role of postapproval study in this process? What kinds of postapproval clinical information are mandated as part of the FDA approval of a new agent?
Gerald Dal Pan, MD, MHS
Dr Dal Pan: We do take a lifecycle approach to a drug product—from the first human exposure all the way through its continued marketing life—and we monitor the safety throughout. We have a particularly active postmarket program and obtain that postmarket safety data in a few different ways. The backbone of our system for over 40 years is adverse event reporting. These are reports describing an adverse event that took place while using a medicine which people at the point of care send either to the drug’s manufacturer or to the FDA using our MedWatch system. These reports, importantly, include untoward occurrences that may or may not be later determined to be related to the drug.
Manufacturers are under a regulatory obligation to send certain reports to the FDA. We receive about 1.7 million reports a year and now have over 14 million reports in our database. About 95% of our reports come from industry and only about 5% come directly from the public—patients, their families, physicians, pharmacists, nurses, anyone. None of those people at the point of care are required to send the report either to the FDA or to the manufacturer. But to clarify: All reports, whether they come to us from industry or come directly to us, are from the point of care.
We use data mining and other computer-based techniques to sift through these reports to identify a signal that suggests that there might be a new problem with the medicine that we want to evaluate. We begin by actually reading the reports to see if there is sufficient information to determine whether a causal relationship exists between the medicine and the adverse event. Often there is not enough information because people who take medicines are experiencing a lot of health problems before they take the medicine. That’s why they take the medicine. It’s hard to separate the effects of the underlying illness and the medicine, so further analysis is necessary.
We can further explore this potential association between the medicine and the adverse event in a few ways. First, we can examine existing observational studies or conduct our own observational study to see what happens in the course of normal clinical care, and then try to make an inference about a causal, or potentially causal, relationship between a drug and an adverse event. We can also ask the drug manufacturer to do a study or a clinical trial to try to address this issue. Another method is to work closely with our clinical pharmacologists here at the FDA to analyze the pharmacology of the medicine and see if there’s a biologic or pharmacologic basis for the drug causing the adverse event. We have a lot of tools in our toolbox.
When certain regulatory requirements are met, the FDA has the authority to require companies to perform a postmarket study or clinical trial. Before we can invoke that authority, however, the law requires us to demonstrate that our active risk identification and analysis system isn’t sufficient to answer the particular safety question. That system was mandated by the Food and Drug Administration Amendments Act (FDAAA) of 2007, and it required FDA to amass data on at least 100 million lives (generally, electronic healthcare data) to be able to answer these questions.
Medscape: So, the FDA can mandate postmarket research to be carried out by the manufacturer after demonstrating that analysis of information from the adverse events reporting system is not sufficient to answer any outstanding clinical questions. This is distinct from postmarket research, via either a randomized clinical trial or an observational trial, which can be mandated at the time of approval. Is that correct?
Dr Dal Pan: Correct. The FDA has several different authorities through which it can mandate postmarketing studies or trials when a concern arises about safety issues. We may identify a safety issue during development and, at the time of approval, require a postmarketing study or clinical trial to clarify further that issue. Alternatively, we can invoke that authority after approval if we obtain new safety information that requires more evaluation.
The general framework for the conditions under which we can require this is either when there is a known safety issue and we need some more information to understand it better, or when a signal of a potential safety issue is identified that we want to understand better. A postmarket study can also be mandated when there is a broader potential for a problem even though we’ve not had any clinical data that might warrant study. An example of the third category would be in the situation where there is a concern that a drug may pose a teratogenic risk and a pregnancy registry is mandated.
There are separate authorities under the accelerated approval regulations, the Pediatric Research Equity Act, and the animal rule, through which the FDA can require a postmarket study. These are generally different from the postmarket safety studies described above and often deal with questions of efficacy.
Medscape: Does the FDA determine the design of studies that will be conducted to generate postapproval evidence?
Dr Dal Pan: We are required to justify what we mandate, be it before approval or after approval. And that is true whether the postmarket research will be conducted via an observational study or a clinical trial. The details of these trials, such as the specifics of the protocol, might not be known at the time of approval. But in either case, whether the study is mandated before approval or after approval, we do work with the pharmaceutical company on the design of the study or trial.
Maintaining Vigilance in the Postmarket Period
Medscape: The FDA’s Sentinel Initiative is a national electronic monitoring system used to identify postmarket risk. Can you describe this initiative? What are the benefits and limitations of Sentinel? What is the role of prescribers in this system?
Dr Dal Pan: Sentinel is one of our bigger drug safety initiatives and was begun in 2007 following the passage of the FDAAA, which required us to put together a system with data on 25 million lives by 2010 and 100 million lives by 2012. We met and exceeded those goals. During the initial work between 2009 and 2014, we launched the mini-Sentinel pilot through a contract with the Harvard Pilgrim Health Care Research Institute in Boston. They then subcontracted with multiple other data partners. These organizations are mainly large health insurers that have significant amounts of electronic healthcare data on defined populations. These data include medical diagnoses and information on dispensed prescriptions, including when a medication was started, and things that might have happened with that patient afterward.
This effort required that the partners develop a common data model for all necessary information so that it is all essentially written in the same language. When a drug safety question arises, we determine whether Sentinel might be an appropriate tool to answer the question. We then form a team with the operating center in Boston, discuss the question, and specify exactly what the query will look like. This way, we have a common understanding of the question and we know what data are needed to provide an answer. The operating center writes a statistical, analytical program using the language of this common data model to share with these data partners. The data partners run the data behind their own firewalls. It’s an important part of the system that each data partner retain control of its data and that individual patient-level data do not go outside of the data partner’s firewall.
The data partners then generate summary statistics which are returned to the Sentinel operating center in Boston. The Sentinel operating system aggregates those data and shares them with FDA for interpretation and analysis.
The Sentinel system does have some limitations. Like all data based on prescriptions claims, it doesn’t include over-the-counter (OTC) medicines because they are not included in prescription databases. It doesn’t have a linkage to mortality records, although we’re working on adding that capability. Because it deals with primarily a privately insured population, it has underrepresentation of people over 65. To get around that, we’re working to integrate the Centers for Medicare & Medicaid Services into that so that the system will include Medicare data.
One feature of postmarket safety is having systems to detect the unexpected. MedWatch is the system that allows us to do that.
Medscape: Sentinel, then, is a passive system in that it’s collecting data that already exist and doesn’t require active reporting on the part of the public. It sounds like the MedWatch and Sentinel initiatives complement each other, but Sentinel does not supplant the MedWatch system—correct?
Dr Dal Pan: That’s exactly right. We’ve always said that Sentinel is another tool in the toolbox. One feature of postmarket safety is having systems to detect the unexpected. MedWatch is the system that allows us to do that. Right now, our use of Sentinel is limited to evaluation of signals of safety. It does not generate signals. These are complementary approaches and we don’t expect Sentinel—or anything else, for that matter—to supplant the MedWatch program. Slightly over half of the safety reports that we generate are the result of adverse event reports.
Getting the Word Out
Medscape: New information gleaned from postmarket research can be conveyed to clinicians and the public in a variety of ways. Can you describe how the decision about the method for disseminating new information, including timeline, is made?
Dr Dal Pan: The primary way that we let practitioners know about new safety issues is through changes to the product label. The same law that allowed us to require manufacturers to conduct certain studies also gave us the authority to require manufacturers to make changes to their label when we have new safety information. So the label has been and still is the primary tool to convey information that’s essential to prescribing the medicine. We may also issue a drug safety communication linked to one of these label changes.
However, sometimes we may issue a drug safety communication before we have the whole story finalized. We may do this to let the public know that there’s a signal that we’re still evaluating.
We also have a MedWatch listserve that sends information to anyone who signs up, be it an organization or an individual, about label changes we make. Professional societies pick up messages that are relevant to the particular specialty society and disseminate the information via newsletters, conferences, etc.
Along with Aaron Kesselheim and his colleagues at Harvard Medical School, we are studying how these messages are disseminated and where people get information. In 2013, we issued a safety alert about zolpidem-containing medicines to warn about next-day drowsiness and impairment, with a recommendation to change the starting dose. We then followed how these changes were distributed through mass media, such as newspapers and social media, and how prescribing changed. We’re interested in better understanding the impact of our actions. Change doesn’t happen overnight, but this new information does get integrated into clinical practice over time.
Medscape: What is your advice for clinicians who are using these newly approved agents in their patient care?
Dr Dal Pan: We encourage adverse event reporting of suspected adverse drug reactions. That report doesn’t have to be long but ideally is a reasonably detailed summary with sufficient pertinent clinical data to allow an independent health professional reading this report—someone not involved in the patient’s care—to make an independent assessment as to what the relationship between the drug and the adverse event might be. So, including pertinent past medical history, reasons for taking the medicine, important diagnostic tests, and the outcome after the event happened is essential.
Beyond that, I would encourage people to read our labels and any updates that we may have on the MedWatch listserve. It’s important for health professionals to educate the general public that all medicines—whether prescription or OTC meds, even ones for which we have decades of experience or brand-new ones—have risks and benefits. The critical decision when prescribing medicines is to determine how that risk-benefit profile relates to the patient I’m seeing now. The information on the label is really designed to help clinicians make that determination.
We know that new medicines are going to have new safety issues identified after their approval. Once approved and in use, these medicines are often prescribed for broader populations than the population included in the clinical trials.
Medscape: In light of these new studies, will the FDA change its postmarket safety surveillance programs in an effort to help alleviate concerns about not-yet-recognized safety risks with newly approved agents?
Dr Dal Pan: Let’s emphasize that this is not a newly recognized concern.We know that new medicines are going to have new safety issues identified after their approval. Once approved and in use, these medicines are often prescribed for broader populations than the population included in the clinical trials. Patients in the real world are on a lot of different medicines concomitantly and may have a variety of illnesses. Medicines may not be used in a way that the label recommends. The safety of medicines is a combination of the inherent pharmacologic properties of the medicine as well as how the medicine is used in our healthcare system.
We have these systems set up to add postmarket safety information because we know this will happen. We know that medicines will have postmarket safety changes that do, in fact, occur throughout the lifecycle. In a study of safety events that resulted in a label change in 2010, half of the medicines that had a label change that year were on the market for less than 11 years and half were on the market for more than 11 years. This really does occur throughout the product lifecycle, and that’s why it’s important for the entire system to be attentive.
Medscape: In the JAMA study, it was unclear whether the safety events had occurred with on-label or off-label use.
Dr Dal Pan: Right. Frankly, that is very difficult to understand in all cases—exactly the conditions of use under which the medicine was used. That’s a tough thing to get at.
Which is why the postmarket safety surveillance system is not a static system but rather one that is always evolving. We’re always looking at new capabilities and methods, as well as our own internal processes—to improve them as well. It is a real, dynamic system, and that is essential.