If pharmacovigilance were a movie, it would be Groundhog Day, except as a tragedy.
Here's the plot. Reports of adverse events (AE) pop up and take years to be evaluated. The associated recriminations bounce between various interested parties. Good drugs exit the market. Bad ones linger. The industry's reputation gets bludgeoned either way. Luck is at work, but doesn't affect the interminable cycle of suspicion, accusation, investigation and compromise.
It's not a news flash. The current drug safety system is not serving nervous patients, litigation-weary drug makers or regulators swamped by individual case safety reports (ICSR). The current regulatory environment typically generates miniscule quantities of usable data and magnificently complex epidemiological goose chases. Spurious allegations can be triggered by news accounts; legitimate safety signals can be too faint to hear.
Worst of all, some adverse events are never submitted to regulators at all. Unsubmitted AE reports are locked inside the minds of tired physicians overburdened by ancient computers. It may take 30-45 minutes to fill out a MedWatch form by hand. Then some poor soul at the FDA retypes it. After all that, there may not be sufficient information for an FDA epidemiologist or an industry-based physician to follow up intelligently.
Pfizer is trying to figure out something a bit more sensible. Michael Ibara is the firm's head of pharmacovigilance information management and designed the Aster project to explore an efficient and clinically acceptable work flow to capture adverse events reports. "We're missing important information that's out there. We've known that for a while," Ibara says. Ibara is speaking at next month's Post-Approval Summit. The meeting is organized by Outcome and is one of the most substantive conferences of the year, with unusual numbers of high-level speakers from industry and the agency.
Besides the fax and paper reports, a new tsunami of digital health care data is poised to swamp safety professionals, Ibara says. He decided to be proactive and reflect on how to completely rethink the situation. Not tweak it. "We cannot simply continue to process paper or faxes or even the way we do electronic reports and expect not to be overwhelmed," Ibara says. "We need to look at it at a system level [and] understand that we have new sources coming in."
With a modest investment, Pfizer decided to try something new in Boston, both in community practice and at Harvard. The Aster effort uses the system the doctors already know—the electronic health record (EHR)—to feed data into a drug safety database that forwards reports to the FDA. Electronically. The project worked on drug safety in post-marketing.
One early insight was that the computer could handle the drudgery of filling out demographic data and other information already in the system. That would leave the physician to add a few key drug safety details.
Less Than A Minute
The Aster physicians were clocked at needing just a minute to fill out the AE form—an impressive improvement. In some cases, just 20 or 30 seconds were needed to generate a huge quantity of high-quality data. Doctors who had never submitted AE forms started doing so.
Again, that's because doctors didn't have to retype a lot of demographic or miscellaneous data. In a post-marketing context, Ibara says, there are no regulatory issues around capturing such information from an EHR system. "We typically don't have an issue of not collecting enough information," Ibara says. "We actually have more information in the EHR than we ask for on a MedWatch form."
Pfizer spent some time sweating the details on the electronic form. Some of the physicians wanted to see everything on a single view. But that would be an eyeful. So Pfizer and its partners decided to have collapsible sections on the form that would allow physicians to open or close discrete sections of the document to a) see everything going to the FDA or b) just review the most important document sections.
Says Ibara: "They can click on those sections and it will expand and they can basically see everything that is going to go to the FDA as an adverse event report. They can also collapse it down. We make it clear that all of this information is going to FDA and they can review all of that if they'd like."
Pfizer knew that big EHR system suppliers would not want to learn the nuances of industry research. Or break their EHR systems for a few physicians who do 21 CFR Part 11-regulated science. So all the Aster technology does is feed information to another server. "The hit, as it were, on the EHR system is to collect the data and pass some and receive some back," Ibara says. "They don't do any core code changes."
The system was designed to do triage on the data. One key sorting mechanism: when physicians were discontinuing a drug.
In those cases, Pfizer felt it would be dealing with a bona fide adverse event related to the usage of medication—not, perhaps, a random backache that had no link to a drug. In the future, the system could be tuned to expedite some data and deprioritize other reports. Says Ibara: "If you do that and you have the interaction of the physician involved, I think you get much higher quality reports."
Managing the Deluge
Indeed, the sheer quantity of adverse events and serious adverse events can be a touchy subject for an industry relying on fax machines and kludgy federal websites. For his part, on the Aster project, Ibara wanted to make sure that followup calls to physicians were as warranted and productive as possible. "I very quickly saw how you could overwhelm yourself with spurious reports," he says, having worked at Quintiles as well.
Ibara readily allows that the quality of drug safety reporting is another facet of the debate over how to modernize the existing system. He tried to come up with a system that produced more useful reports. The goal: to let the scientific chips fall where they may, but help clinicians and regulators delve into the risks of the drug at hand as quickly as possible. "The discussion revolves around the question, will we get in more events or more information that are the same or higher quality? And I think the answer to that is, it depends. It depends on the kind of system you design," he says.
On The Horizon
It's clear that Pfizer and other large pharmas are struggling to keep up with hundreds of thousands of AE reports. Given that, it's hard not to wonder if the federal government has similar problems. Does the FDA have the needed human, technology and financial resources? Will the U.S. Sentinel effort handle the coming tide of electronic safety data? Ibara is enigmatic on those questions, but sounds worried that a moment of truth may not be far away. There's just too much information. "With that volume of information, we can no longer process events the way that we do now," he warns.
"For a long time, folks in the industry will know, a lot of us have had this sort of one-to-one correlation between manpower and number of events," Ibara says. "And that simply will be impossible to maintain, if it's not already. If you agree that that's impossible to maintain, then you have to start asking yourself, how am I going to use technology to help me process this information? That applies to FDA as well."
Ibara says that industry is already confronting the limits of the present approach. Screening or sorting adverse events automatically could help, he suggests. That would handle the most immediate issues. It's not clear if politicians could concede, in public, that industry has unique expertise and resources to do that.
Some sort of categorizing of AE reports, Ibara notes, would not preclude more conventional data mining or careful, rigorous research on the entire dataset at a later time. Ibara puts it this way: "What you need to accept in your assumptions, that hasn't been true before, is that we can't treat everything with the same priority as it comes in now. In fact, you know, our processes [at Pfizer] don't separate between non-serious and serious events, but we can't process everything the same. It will simply overwhelm us."
What the industry needs, he suggests, is a new way of working with the reports. "We're talking about a new business model here. It needs to change for the reporters, for the doctors reporting the adverse events, it needs to change, to some degree, for the industry and it needs to change for the regulators. And after, you know, being depressed for a few days over that issue, I thought, well, if this is what it is, this is where we need to go."
Another story on our site has some of Ibara's ideas on how the industry and regulators could move forward. A new podcast is free but requires registration. That page includes a 15-page transcript of the interview.d9A2t49mkex