The FDAs Sentinel Initiative
Many prescription drug risks only become evident after they reach the market and begin being used.
What’s the issue?
Prescription drugs are a mainstay of medical care. In particular, people who have chronic conditions such as high blood pressure, diabetes, high cholesterol, or arthritis may take prescription medicines every day for years or decades. The majority of Medicare beneficiaries take at least one medicine for a chronic condition, and 25 percent take five or more. Generally, these treatments have been shown to manage or improve patients’ health. But all drugs carry some risks. Sometimes these risks emerge in studies of the drug before its approval and are known by regulators, drug makers, and health care providers when a drug hits the market. Many risks, however, only become evident after drugs reach the market and begin being used.
For example, the nonsteroidal anti-inflammatory drug rofecoxib (Vioxx) was approved by the Food and Drug Administration (FDA) in 1999 and widely promoted for use as a pain reliever. But studies and analyses done after Vioxx hit the market confirmed that it raised the risk of heart attack. An estimated 40,000 people died as a direct or indirect result of taking the drug before it was taken off the market in 2004. The diet drug fenfluraminephentermine (Fen-Phen) also was removed from the market after studies found its risks outweighed its benefits and that people had been harmed. Thus, vigilance about the safety of approved drugs is essential.
Until recently, however, the FDA had primarily relied on passive collection of adverse events obtained from manufacturers or through voluntary physician and consumer reporting to track emerging safety issues after drugs received FDA approval. The FDA also learns about potential safety problems from published studies and—on rare occasions— from studies conducted by drug makers to address a specific question posed by the FDA.
There are inherent limitations to relying both on passive collection of adverse event reports and on postmarket studies conducted by drug makers. For example, spontaneous reports may lack complete data and have little power to detect common events such as heart attacks and strokes that might be increased in connection with taking a drug. Prospective trials of safety issues are costly and can take years to conduct.
A confluence of recent developments—electronic health records (EHRs); new software tools; the advent of big-data analytics; and the increased focus on coordination and alignment among health systems, doctors, payers, and government to solve systemic problems in care delivery—now makes it possible to morequickly and more efficiently track patient outcomes and treatment results on a larger scale.
One leading example of how these developments are changing the health care system is the FDA’s Sentinel Initiative, which aims to use big data and broad networks to proactively and systematically detect and respond to emerging risks associated with prescription medicines.
The FDA and Harvard Pilgrim plan to increase the number and types of Sentinel data partners over the next few years. At the same time, FDA officials have pledged to collaborate on the creation of a national data infrastructure that would “enable other users (for example, other governmental agencies, researchers from academia or industry) to access the Sentinel infrastructure for medical product research and quality improvement.”
To achieve that goal, the FDA says it will partner with other data initiatives. Those include the Innovation in Medical Evidence Development and Surveillance program, now housed at the Reagan-Udall Foundation, an organization established by the FDA Amendments Act of 2007; the National Patient- Centered Clinical Research Network housed at the Patient-Centered Outcomes Research Institute; and the National Institutes of Health Collaboratory Distributed Research Network.
Sentinel is at the cutting edge of the big-data revolution. It has tremendous promise. But the FDA has yet to prove that it can use this new tool to its full potential. That may now change as Sentinel transitions from a pilot project to a full-scale program.