Sentinel to Transform Drug Development and Oversight

The US Food and Drug Administration has launched the Sentinel Initiative to expand and improve medical-product risk identification and evaluation. This program aims to implement a provision in the FDA Amendments Act of 2007 (FDAAA) that requires the agency to establish an "active postmarket safety surveillance and analysis" system. Instead of depending on passive, unreliable adverse-event reports sent to FDA by doctors and pharmacists, the agency should detect safety signals much more quickly through links to databases that contain health information on millions of patients.

Jill Wechsler

FDAAA instructs FDA to establish a system that accesses multiple information sources, providing safety data on 25 million patients by 2010 and 100 million by 2012. This expanded drug-safety assessment program, dubbed the Sentinel System, also will collect more adverse-event (AE) reports from health professionals, provide effective analysis of health information, and enhance FDA methods for communicating new safety information to providers and patients. "We may be entering a new era for determining how drugs should be used on the market," commented Mark McClellan, former FDA commissioner and Medicare chief, now at the Brookings Institution, where he chaired a June 2008 forum on postmarket evidence.

JASON REED/GETTY IMAGES

This shift to active postmarket surveillance promises to have a broad impact on biomedical research and drug development; an improved understanding of how drugs affect patients in real-world settings can inform drug discovery. This approach creates an opportunity to use the delivery of healthcare as a discovery platform, said FDA Commissioner Andrew von Eschenbach at the Brookings forum. Garry Neil, Johnson and Johnson's corporate vice-president, predicted that this development can "revolutionize the entire healthcare system." Bringing information about how drugs affect people back to discovery, he explained, will support new medical products that are more effective, safer, and less costly than older products.

In Washington This Month

McClellan further explained at the June 2008 annual meeting of the Drug Information Association (DIA) that an active postmarket surveillance system reflects a move away from distinct pre-and postmarket regulatory processes and toward a life-cycle approach to drug development and regulation. Rising FDA concerns about drug safety are "not just a pendulum swing," he said, but reflect higher public expectations about drug safety over the longterm.

Modernizing AERS

The development of a fully active postmarket surveillance system is still years away. FDA is "just getting started" on this major epidemiological piece of FDAAA, said Sandra Kweder, deputy director of the Center for Drug Evaluation and Research's (CDER) Office of New Drugs. Meanwhile, FDA is implementing the Sentinel System in stages, building on existing systems and public–private partnerships that can answer questions about apparent drug side effects.

An important step is to modernize and expand the agency's current spontaneous adverse-event reporting system (AERS). The system collects AE reports from health professionals and the public, who voluntarily file information on drug-related medical problems. Although passive reporting fails to capture many AEs or explain the cause of medical events or the extent of exposure, it is useful for detecting signals and for developing hypotheses about rare but serious safety issues.

FDA is working to improve methods for assessing the reports it receives through this system and to boost voluntary reporting. In March 2008, the agency issued a five-year drug-safety plan, as required by the newly revised drug user-fee program (PDUFA IV). The plan describes how the agency will improve the timeliness and accuracy of AE data. To boost voluntary reporting, FDA is collaborating with the Centers for Disease Control and Prevention (CDC) and the National Institutes of Health (NIH) on standards and a common format for reporting AEs for all FDA-regulated products. This approach will make it possible to link AE information from clinical studies with that from postmarket surveillance and will facilitate the aggregation and analysis of AE information.

FDA also plans to combine safety-signal detection and analysis for drugs, medical devices and other regulated products through an agency-wide FDA adverse-event reporting system (FAERS). And a user-friendly MedWatch Plus portal will allow anyone (e.g., healthcare providers, patients, and clinical researchers) to electronically report adverse events on all regulated medical products.

Although still voluntary, these initiatives will help FDA collect timely, accurate AE data, pointed out Armando Oliva, deputy director for bioinformatics in FDA's Office of Critical Path Programs, at the DIA annual meeting. More standardized data and better analytical tools will enable FDA to improve safety assessments. "Spontaneous reporting won't vanish," said Janet Woodcock, director of CDER, at the Brookings forum. But it will be conducted more efficiently than before and linked to broader analytical resources.

Distributed network

This expansion of AERS, however, is not designed to build a giant FDA health-information database for medical-product monitoring. Instead, the agency will expand postmarket surveillance by accessing data on drug use, patient outcomes, and specific safety issues held by private health plans, insurance companies, database operators, government agencies, and medical-products manufacturers. The aim is to negotiate contracts and partnerships with these existing electronic databases to establish a national, integrated, electronic distributed network for monitoring medical-product safety. By using a distributed network, all the personal medical data remains behind firewalls, thus reducing concerns that FDA is compromising patient privacy and data security.

FDA has been laying the foundation for the Sentinel System for years through arrangements with healthcare "data owners" that permit it to pose safety-specific questions for evaluation through electronic health records, claims databases, and other information sources. For example, the agency has contracted for several years with UnitedHealth Group's Ingenix, Vanderbilt University, Kaiser Foundation Research Institute, and Harvard Pilgrim Health Care to access pharmacoepidemiologic information that can test hypotheses arising from AE reports. FDA's Center for Biologics Evaluation and Research (CBER) works with CDC, the Veterans Health Administration (VHA), the Centers for Medicare and Medicaid Services (CMS), and large health plans to monitor influenza-vaccine use and related outcomes. FDA also supports the Severe Adverse Events consortium to gain more understanding of the genetic basis of drug-induced serious AEs. Several projects with the Agency for Healthcare Research and Quality aim to assess the use of distributed networks for comparative effectiveness and safety studies.

A new CMS policy now gives FDA access to one billion pharmacy claims per year, collected by the Medicare Part D drug-benefit program, a treasure trove of healthcare-use data (see sidebar "Evaluating Part D"). With additional funds to support postmarket safety surveillance, FDA plans to expand its access to CMS, VHA, and health-plan databases to further expand its ability to detect and assess AEs related to marketed drugs.

Evaluating Part D

Shaping the system

FDA and its partners are beginning to map plans for operating the distributed network, but the details are still "a matter of discussion," said Gerald Dal Pan, director of CDER's Office of Surveillance and Epidemiology, at the DIA annual meeting. FDA, health plans, and database operators expect to establish a charter that defines which entities may join the partnership, what data sources will be used, what queries can be run on the system, who will be able to run queries, how specific evaluations will be paid for, and how privacy will be protected. The idea is for a nonprofit network "convener" such as the Reagan–Udall Foundation or an academic research entity to channel information and queries to the partnership and determine how results will be communicated to network participants and to the general public.

The partnership also will support research into best practices and analytical methods and will establish standards for postmarket assessment, most likely under the direction of a scientific advisory board. Improved statistical methods are needed to establish causality and to distinguish meaningful signals when working with large databases. New models for data aggregation that protect patient privacy will be important for enabling FDA to trace back to individual cases when necessary.

Many of these issues will be examined by an active-surveillance pilot project sponsored by the Pharmaceutical Research and Manufacturers of America (PhRMA). After more than a year in development, the project is being launched as a public–private partnership involving PhRMA, FDA, and the Foundation for NIH, with Woodcock as chair. The aim is to test approaches for organizing and operating a drug-safety query system using a distributed information network. The pilot will be transparent, and all parties—industry, academia, and FDA—will have access to its findings.

Streamlining research

An important promise of an efficient and active postmarket surveillance system is that it will support FDA approval of more new drug applications. If the regulators can expect that safety problems will be identified quickly and appropriately after a drug goes to market, they may be more willing than before to approve a market application for a product that raises some risk concerns. The hope is that the Sentinel System will give the public and FDA officials more confidence about approving innovative and targeted medical products.

A related benefit of effective surveillance similarly could be to reduce the time and expense of pharmaceutical research and development. Phase III studies are large and costly because they are needed to address safety issues about drugs, McClellan points out. But a robust system for monitoring how drugs are used by patients, he says, "takes off some of the pressure on Phase III studies to address all the safety issues before a drug comes to market."

Pharmaceutical marketers are nervous that earlier and broader assessment of postmarket data will yield false negatives that set off alarms on every new drug. But FDA officials believe the Sentinel System may reduce the need for some postmarketing studies as health-system information demonstrates its ability to identify safety concerns. Expanded postmarket assessment may allay concerns about the safety of a new drug or vaccine that are raised in prelicensure studies, said CBER Director Jesse Goodman: "The data can cut both ways." Instead of requiring a manufacturer to sponsor a long list of postmarket clinical studies, FDA reviewers may call for six months or a year of database monitoring for certain anticipated safety concerns.

In the future, it may be possible to monitor medical-product safety through a Nationwide Health Information Network that connects clinicians, consumers, and providers across the healthcare system. Government agencies such as FDA would have access to data for public-health purposes, but would have to address privacy issues and ensure that health-information exchanges are secure. FDAAA instructs FDA to examine methods for ensuring the privacy, confidentiality, and security of patient information in establishing an active postmarketing risk-identification and analysis system, an issue that the Government Accountability Office will report on next year.

A linked system would provide a broader picture of a health problem and improve tracking of patients that shift to different healthcare providers, Woodcock noted at the Brookings forum, but it also would require a higher level of consent and transparency. "This could be something for the future," Woodcock said. "We would not do this to start, but don't want to rule it out."

Effective communication about safety signals from interim results will be a challenge, as will efforts to provide adequate and sustained financial support for what will be a costly undertaking. These are issues that FDA and its Sentinel partners will address, as well as concerns related to scientific accuracy and data reliability, said von Eschenbach at the Brookings forum. He emphasized that it is important not to regard every problem as a "fatal flaw" in an active surveillance system. FDA is not implementing the Sentinel System "solely because Congress told us to, but because it's the right thing to do," he asserted. "FDA must, will, and needs to do this."

Jill Wechsler is Pharmaceutical Technology's Washington editor, 7715 Rocton Ave., Chevy Chase, MD 20815, tel. 301.656.4634, jwechsler@advanstar.com