Mini-Sentinel Expanding Drug Safety Efforts To Prospective Surveillance

The effort will complement the FDA-funded project’s existing ability to retrospectively measure exposure to products and then analyze specific adverse outcomes.

FDA’s Mini-Sentinel post-market drug safety surveillance program is developing new capabilities to prospectively monitor new products entering the market on a routine basis in near real time. The effort will complement the project’s existing ability to retrospectively measure exposure to products and then analyze specific adverse outcomes.

Now in the fourth year of a five-year contract, Mini-Sentinel is a pilot project for FDA’s Sentinel Initiative envisioned as a national active post-market safety surveillance project drawing from medical claims data and electronic health records provided by payers and health care providers. Mini-Sentinel currently has access to 380 million “person years” of data provided by 17 partners (see box).

FDA’s Patrick Archdeacon discussed plans to expand the scope of Mini-Sentinel’s work during the annual Sentinel Public Workshop in Washington, D.C., Jan. 31, convened by the Engelberg Center for Health Reform at Brookings. Archdeacon is a medical officer in the Office of Medical Policy within FDA’s Center for Drug Evaluation and Research.

“Right now [Mini-Sentinel] has a pretty good infrastructure in place that does a great job of capturing exposures to the various [drug] products that come to market,” Archdeacon said. The program also is “very good at looking at a handful of outcomes,” including acute myocardial infarction, bleeding and thromboembolism, and “that allows us to look at these exposures for those outcomes.”

“The difference here is that if we can get this semi-automatic system, we’re going to be able to identify the outcomes that we should be concerned about, and then we can bring our resources to bear” – CDER’s Archdeacon.

“We’d like to move to a place where … we increase those capabilities but also we want to vastly expand our abilities so in near real time, we’re able to look across a range of products at a range of outcomes,” he said. “We need to develop expertise at looking at those outcomes,” but “once we do that, we’ll be closer to a comprehensive system.”

Archdeacon acknowledged a prospective system will have less ability to “control for systematic biases” in the data, but explained it is expected mainly to help the agency generate “hypotheses” on possible connections between new drugs and adverse events. “One of the issues with Mini-Sentinel is that right now we have to already have developed the outcome we are concerned about,” he pointed out. Outcomes of concern so far have arisen from the agency’s passive adverse event reporting system and the safety literature.

“The difference here is that if we can get this semi-automatic system, we’re going to be able to identify the outcomes that we should be concerned about, and then we can bring our resources to bear,” Archdeacon said. He noted the prospective surveillance system also would be “less resource intensive” than the existing retrospective program.

The prospective system is in the early stages of development. Efforts under way are focused on vastly expanding the “menu” of measurable outcomes that would be tracked. Sebastian Schneeweiss of Harvard Medical School is involved in the effort and discussed his perspective on some of the other challenges ahead. One priority will be “coming to some sort of agreement as to what sort of risk level is acceptable, because that will inform the learning later on,” he pointed out. “If you want to automate things, you have to make these decisions up front.”

Schneeweiss is principal investigator of the Brigham and Women’s Hospital DEcIDE Research Center on Comparative Effectiveness Research and the DEcIDE Methods Center, both funded by the Agency for Healthcare Research and Quality, and director of the Harvard-Brigham Drug Safety Research Center, funded by FDA.

He said preparatory considerations for a prospective system also include “when does it make sense to set up a monitoring system” to collect experiences with a new drug, “and related to that, when should we stop monitoring, when do we know enough?” The decision “comes down to thresholds, what is the acceptable level of risk,” he suggested. “If you are long enough within that level of risk then you may want to conclude at that point in time the drug seems to be safe.”

Archdeacon said FDA is not planning to consult with drug sponsors on specific products at this stage of developing the system. “We’re in the first generation phase where we’re developing the tools,” he pointed out. “We haven’t anticipated engaging with sponsors on specific products at this point.” He also noted that “the whole point of doing prospective surveillance is that we’re going to be casting these tools across a wide range of products simultaneously in a way that isn’t going to require fine tailoring to a product exposure outcome, where a sponsor may have a particular expertise to bear.”

On the other hand, he said, “if we got to a point where we generated a signal or generated a hypothesis and we wanted to take it to the next stage of doing a protocol-based assessment, I think we might want to consider at that point” some engagement with sponsors. Archdeacon emphasized, however, that the new system would function as “just one source of information, and there are many others.”

Drug Industry Concerns

The plan has caused concern among drug industry stakeholders worried about the false positive signals that might be produced by the system and the negative impact that could have on new products. GlaxoSmithKline Senior VP for Clinical Evaluation Sciences Ralph Horowitz suggested at the workshop that Mini-Sentinel studies to date provide some basis for concern.

Specifically, he said, the use of unadjusted risk estimates in the Mini-Sentinel study on bleeding risk for Boehringer Ingelheim Corp.’s Pradaxa (dabigatran) versus warfarin “leaves a little bit of anxiety” because clinical considerations were not taken into account and might have an important bearing on outcomes. “Not to … adjust the risk estimates for differences in clinical features that are likely to be associated with the choice of dabigatran or warfarin – you know you can get an unadjusted estimate, but what will it mean?” he asked. The Mini-Sentinel comparison was among the key data used in FDA’s recently announced decision not to change its recommendations for use of Pradaxa following reports of severe bleeding events (Also see "Mini Sentinel Alleviates FDA Concern Over Pradaxa Bleeding Issue" - Pink Sheet, 2 Nov, 2012.).

Horowitz said, “We need to come to terms with how this routine pharmacosurveillance takes place and the kind of information we need to have confidence about the inferences that will be drawn from it. I know it is part of a larger database that ultimately leads to decisions, but data, especially precise data, tend to be highly influential, and there’s nothing worse than precisely wrong data.”

Mini-Sentinel’s work in responding to FDA queries on possible drug-event connections has so far included what FDA calls standard reviews and more resource-intensive protocol-based evaluations.

Mini-Sentinel By The Numbers

• 380 million person-years of data
• 17 data partners
• 400 investigators affiliated with partner organizations
• 307 FDA queries run to date
• $10 million a year to maintain data set and run standard queries
• $225,000 to produce a protocol-based study

Source: Presentation by Mini-Sentinel Principal Investigator Richard Platt at Jan. 31 Sentinel workshop.

The standard reviews are designed to rapidly respond to an FDA request seeking drug exposure and adverse outcomes information for a specific subset of individuals over a specified period of time. Protocol-based evaluations are longer-term efforts. Current protocol-based evaluations now in progress include an assessment of the impact of FDA-required labeling changes on use of long-acting beta agonists for asthma (Also see "Mini-Sentinel Project Will Evaluate Impact Of LABA Label Change" - Pink Sheet, 26 Sep, 2011.). A draft final report on that study is scheduled to be completed in April.

Protocol studies in progress also include investigations into the connections between anti-diabetic drugs such as Bristol-Myers Squibb Co./AstraZeneca PLC’s Onglyza (saxagliptin) and acute myocardial infarction, as well as rotavirus vaccine and intussusceptions and Merck & Co. Inc.’s human papillomavirus vaccine Gardasil and thromboembolism (Also see "Onglyza Data From Sentinel Pilot Can Be Validated By Phase IV Trial, FDA Says" - Pink Sheet, 17 Jan, 2011.).

Studies with protocols under development include analyses of the metabolic effects of antipsychotics in children and adolescents, the connection between Pradaxa and stroke, influenza vaccine and febrile seizures, influenza vaccine and birth defects/spontaneous abortion, intravenous iron products and anaphylactic reactions, I.V. immune globulins and thromboembolic events and a general influenza vaccine safety analysis.