A decade after the US Food and Drug Administration (FDA) launched its Sentinel system for active postmarket surveillance of medical products, officials from the agency say that new technologies such as natural language processing and machine learning will be key to expanding its use going forward.
Speaking at the Duke Margolis Center for Health Policy's Annual Sentinel Initiative public workshop on Wednesday, FDA Commissioner Scott Gottlieb said that Sentinel now has access to data for more than 223 million US patients, with both medical records and pharmacy benefit information for 178 million.
Gottlieb also said that FDA has conducted 254 analyses using Sentinel since the system became fully operational in 2016, which Harvard Professor and Sentinel Principle Investigator Richard Platt said would have been far too expensive for the agency to perform without Sentinel.
"We don't have even a back of the envelope estimate for what would it have cost to do these analyses using the old kind of system. The answer frankly is more than the agency could afford, so many of these analyses are ones that never would have been done absent this," Platt said.
But despite Sentinel's usefulness for assessing safety information so far, Gottlieb said the agency must begin to consider how to use Sentinel to answer more complex questions about products on the market. FDA also committed to expanding Sentinel's capabilities under the recently reauthorized Prescription Drug User Fee Act (PDUFA VI).
"We should consider how Sentinel might be used to answer questions about efficacy," Gottlieb said, especially in situations when those questions relate to a product's safety.
For instance, Gottlieb said Sentinel could potentially be used to answer lingering questions about the safety and efficacy of long term opioid use.
"This clinical question is something we ought to be able to fully understand. Right now we don't have good data on the long term efficacy of opioid drugs, even though they're prescribed for these purposes," he said, noting that careful analyses of electronic health records (EHRs), insurance claims data and other health information repositories could provide answers.
Gerald Dal Pan, director of FDA's Office of Surveillance and Epidemiology, said that while Sentinel has come a long way since its inception, the system still has a way to go before it can fully harness the information in EHRs to identify new safety issues as the emerge.
"We've not really used Sentinel yet to improve signal detection, we've used it more for signal strengthening and signal analysis, but signal detection is something we want to do," he said.
According to Robert Ball, deputy director of the Office of Surveillance and Epidemiology, the biggest challenge for FDA going forward is dealing with these issues across a distributed network of partners and data sources.
"The ten-year vision is actually very easy … but it's the incremental steps we take to get there that are important," he said. For instance, not all the records Sentinel has access to are machine readable, so natural language processing technologies will be limited to the data that is in a format that can be readily processed.
CBER Update and BEST Contract
On the biologics side, Steven Anderson, director of the Office of Biostatistics and Epidemiology at the Center for Biologics Evaluation and Research (CBER), said that FDA has dramatically increased the use of its Active Postmarket Risk Identification and Analysis (ARIA) system over the last year.
In addition to the 40 queries and 10 protocol--based activities CBER ran through ARIA, Anderson said the center also launched a rapid surveillance system for the annual influenza vaccine.
Anderson also gave an update on CBER's Biologics Effectiveness and Safety (BEST) program, which is meant to expand CBER's Sentinel capabilities by bringing on additional data sources and developing new research tools.
In September 2017, CBER awarded two one-year contracts to IQVIA (formerly Quintiles/IMS) and Observational Health Data Sciences and Informatics (ODHSI) to begin to build out the program. Next week, CBER will also hold an industry information day to answer questions from vendors interested in applying for next fiscal year's BEST contracts.
According to Anderson, CBER is expecting to add some 50 to 70 million EHRs and 160 million patients' insurance claims records through the BEST contract and that the tools being added will allow CBER staff to design queries from their computers from start to finish.
"It's still in the early phases, but we're expecting to gain great efficiencies from this," he said.