FDA drug safety review to start with Medicare
■ The new data-mining initiative could mark an important step toward a system of postmarket surveillance.
By Kevin B. O’Reilly — Posted June 16, 2008
Under intense criticism for its slow response to drug and device safety problems, the Food and Drug Administration last month announced a new initiative to detect adverse event patterns and warn physicians earlier.
The effort, known as the Sentinel Initiative, marks a significant shift from relying on physicians and patients to report adverse events toward a program of high-tech, active postmarket surveillance.
The initiative "will be a quantum leap forward in the FDA's capacity to monitor the use of medical products currently on the market," Health and Human Services Secretary Michael Leavitt said at a news conference. "Before Sentinel, tracking drug safety was like looking at stars with the naked eye -- you could only see what was above your head. Now it's like using a telescope."
The program is still at the testing stage but received a big boost when a final Centers for Medicare & Medicaid Services rule went into effect this month allowing safety researchers to mine deidentified Medicare Part D patient data. The Food and Drug Administration Amendments Act, supported by the AMA and enacted last year, requires the FDA to develop an active risk identification system.
HHS and FDA officials said they hope to access private-sector patient databases as part of the postmarket surveillance effort, but those discussions are just beginning. The Pharmaceutical Research and Manufacturers of America hailed the new initiative, and experts said it could mark a turning point in improving drug safety.
"The availability of this data in this form is probably the most important event in pharmaceutical outcomes research in my lifetime," said David O. Meltzer, MD, PhD, a member of the Institute of Medicine panel that in a 2006 report called on the FDA to improve postmarket surveillance.
Dr. Meltzer warned, however, that technology is no cure-all. Researchers "will unquestionably find associations due to confounding variables or coincidence. The critical question is how well we do in making wise regulatory decisions based on these very noisy and confusing signals."