Applied Clinical Trials
FDA is modernizing systems to better access data on drug effects, utilization, and safety.
Public and private health care experts have been struggling for years to establish electronic health information systems able to track and transfer a broad range of data in patient medical records. The inability of the nation's fragmented health IT (HIT) systems to communicate with one another has made the task extremely difficult, as do concerns about patient privacy and protecting access to personal health information. Now added financial incentives are slated to drive adoption of electronic health record systems (EHRs) by doctors and hospitals, and increased government funding may finally lead to standards for interoperability and eHealth exchanges necessary for the free flow of secure health data.
Increased connectivity among health care entities promises to support more efficient drug testing and development, along with more timely medical product monitoring and oversight. To take advantage of these developments, FDA is working to modernize its internal IT operations for reviewing applications for new drugs and medical products; for receiving and archiving clinical trial data and regulatory submissions; and for tracking manufacturing facilities and product supply chains.
A main goal is to provide more timely information on the safety of regulated products. To this end FDA is centralizing adverse event (AE) analysis through an all-FDA Adverse Event Reporting System (FAERS) and is creating a MedWatch Plus single portal for public AE reporting. Probably the most high-profile project involves establishing a proactive electronic information system that can detect AE signals for medical products more quickly. The FDA Amendments Act of 2007 (FDAAA) requires FDA to augment its passive AE reporting system (AERS) with a more active program able to monitor health records on 100 million people by July 2012.
To fulfill this mandate, FDA launched the Sentinel Initiative in 2008 and now is establishing a "mini-Sentinel" system that will tap into medical records held by large health plans and insurers. FDA can pose queries triggered by evidence of risks seen in clinical trials or by early AE reports to databases established by CIGNA, Kaiser Permanente, and the HMO Research Network, among others. Using this distributed data model, Sentinel partners can monitor for certain events and assess evidence of health conditions associated with drug use. A second phase next year will permit direct FDA access to de-identified patient data from Medicare and other government health programs.
FDA recently signed a $72 million contract with the Harvard Pilgrim Health Care system to coordinate a program that can obtain near real-time signals and ensure data quality. The initiative already has access to information on 60 million patients, according to project director Richard Platt of Harvard Pilgrim and Harvard Medical School, well above the initial FDAAA requirement that FDA monitor 25 million patients by July 1 of this year. Mini-Sentinel also will work with researchers and technical experts at academic centers, private firms, and nonprofit organizations to examine methods for querying against a common data model and to establish policies and standards for validating and analyzing information from diverse information sources.
Platt and others discussed the technical, legal, and administrative issues involved in building the Sentinel System at a January workshop organized by Mark McClellan, former FDA commissioner and now director of the Engelberg Center for Health Care Reform at the Brookings Institution. FDA commissioner Margaret Hamburg opened the conference by noting the growing importance of postmarketing surveillance of regulated products. To have a fully functioning Sentinel System, she noted the need to design a common data model that can compare and analyze data sets, and the need to seek consensus on methodologies for proving and disproving causal relationships between a product and an outcome.
Janet Woodcock, director of the Center for Drug Evaluation and Research, described Sentinel as one element in a broader FDA effort to encourage safe use of medications, to revamp the agency's pharmacovigilance system, and to explore the use of social media to publicize safety issues. FDA also is participating in numerous worldwide collaborations that aim to better understand drug effects in different populations. Judy Racoosin, FDA scientific lead on Sentinel, emphasized that the initial system will help the agency understand the intricacies of gathering safety signals on drugs from multiple sources and closer to real time. Sentinel is "very much a work in progress," Woodcock commented, noting that it will be very different in five years as the science and technology mature.
FDA's decision to launch Sentinel based on a distributed data model reflects earlier research by academic experts and by the Observational Medical Outcomes Partnership (OMOP) that this approach can best ensure the privacy and security of patient health records. The distributed data model avoids FDA building its own mega-database and instead leaves individual patient health information with the insurance company or health system. That data source evaluates the information, transmits summaries to Sentinel, and confirms a specific diagnosis or report if needed. It remains to be seen, though, how well FDA can assess safety signals through this arrangement. Even with the ability to do their own analysis on deidentified data from Medicare, FDA staffers still may find it hard to confirm that a safety issue is real.
Another tricky issue for sponsors and policy makers is the timing of public disclosure of safety information. "The biggest risk to Sentinel is false positives," comments Marcus Wilson, president of HealthCore, the data analysis subsidiary of WellPoint; too early release increases the possibility that a report may be wrong or incomplete and thus raises unnecessary alarm. But delaying communication on an emerging safety issue may result in patient harm and expose program participants to "failure to warn" liability charges. Legal experts want FDA to develop model procedures for when, how, and to whom to report drug safety findings to set a policy that can be applied in court.
All these issues will shape how FDA interprets and utilizes Sentinel findings in regulatory decision-making. Confirmation of safety signals could result in changes in a product's labeling, agency alerts to health care professionals, and public warnings about appropriate drug use. Despite concerns about the completeness, objectivity, and timing of drug safety reports, Woodcock maintains that "what we're going to get is going to be better that what we have right now."
Sponsors want to be at the table when such important issues are discussed and decided and seek to do this through OMOP, an industry-funded collaboration involving FDA, the Foundation for the National Institutes of Health, and the Pharmaceutical Research and Manufacturers of America (PhRMA). Unfortunately, there are no industry representatives on the Sentinel planning board, and OMOP is not an official participant in the program, largely because some stakeholders fear that sponsor involvement would taint the validity of results.
OMOP's agenda reflects strong industry interest in shaping and supporting Sentinel policies for assessing drug safety and for communicating those findings to health professionals and patients. At the Sentinel workshop, Patrick Ryan, manager of drug development sciences at GlaxoSmithKline, described OMOP feasibility studies on data infrastructure and analytical methods completed last year and current plans to evaluate procedures for using observational data to uncover relationships between drug exposure and outcomes.
Pharma companies also want access to Sentinel data for their own internal analyses and for postmarketing studies, and may feel compelled to spend valuable resources building their own versions of Sentinel if blocked from such important information. Woodcock noted that the mini-Sentinel system can't yet support third-party postmarketing analysis, and that FDA will have to deal with industry access once the full "industrial strength Sentinel" emerges.
While Sentinel's prime purpose is to serve FDA's regulatory mission, the long-term vision is that it will be part of a larger health information system that also provides data for outcomes studies, comparative effectiveness research (CER), and health system quality reporting. Manufacturers support the goal of building such a multipurpose network, said Paul Stang of Johnson & Johnson, but this requires more investment in research methods, standards for data sources and evidence, and governance of emerging models.
Carolyn Clancy, director of the Agency for Healthcare Research and Quality (AHRQ), concluded the workshop by describing AHRQ initiatives to improve patient registries and to build distributed data research networks that can answer queries related to the effectiveness and safety of medical services and products. These activities reflect the over-arching goal of identifying synergies between postmarket surveillance and CER, and providing incentives for health care providers to participate in eHealth activities.
Jill Wechsler is the Washington editor of Applied Clinical Trials, (301) 656-4634 email@example.com