Death typically is not included in electronic health records, and sources are disparate; Difficulties with endpoint expose potential limitation to real-world evidence use.
Death, likely among the easiest real-world data points to measure, is proving among the more difficult to gather, illustrating that while real-world data holds great promise, there still are fundamental hurdles to overcome before its widespread use.
Researchers and regulators are hopeful that electronic health records and claims data, so-called real-world data, can be shown to correlate well with clinical trial results and eventually support regulatory decisions. But to get there, they face a surprisingly large hurdle: How to tell when someone is dead.
Accounting for patient death using the records proved difficult during a recent real-world data exercise sponsored by the Friends of Cancer Research because of a lack of reliable and readily available sources. The problem lengthens the time before data sets become available and could complicate the adoption of real-world evidence as a new data source for regulatory use.
“It would really be quite valuable to this entire effort if there were a centralized up-to-date death data repository. Frankly that would drive the field forward in a really remarkable way.” – Andrew Norden, Cota Inc.
Nancy Dreyer, chief scientific officer at IQVIA Real-World and Analytic Solutions, said during the July event presenting the results of the Friends of Cancer Research (FOCR) pilot that patient death is not generally noted in patient records or medical claims.
Dreyer said that while there is a national death index that is available and “phenomenal,” but the data is 12 to 18 months behind. “We need to raise awareness how important those linkages are for efforts like this,” she said. “The system is such that most people aren’t going to call the doctor to say my husband passed away. They’ve got other things on their minds.”
FOCR convened six data partners to explore whether real-world evidence would correlate with clinical trial data in advanced non-small-cell lung cancer (aNSCLC) treated with immune checkpoint inhibitors. The results indicated that several real-world data points correlated with median overall survival measured in the clinical trials, but among the lessons learned from the project was that a more comprehensive and more readily updated database of deaths is needed.
“There was a major effort for years to get physicians to simply note in the record they received a call from a hospice agency that the patient had passed away and compliance with that was extremely poor,” said Andrew Norden, chief medical officer at health care data and analytics firm COTA Inc.. “It would really be quite valuable to this entire effort if there were a centralized up-to-date death data repository. Frankly that would drive the field forward in a really remarkable way.”
Norden said even if there was an incentive in place for physicians to add death to the medical record, the data likely would remain incomplete. He added that policymakers should consider changes.
“I think patients do die in lots of places and I think we would miss a lot,” he said. “I would put this on a central body to make sure death data is captured and shared and done so in a timely way because it benefits us in a lot of different ways.”
The US FDA does not appear ready to make decisions on submission packages with only real-world data. Indeed, in oncology drug development, without a reliable source for death data, FDA may raise questions about the validity of real-world data findings.
The agency wants more groups to attempt to link real-world evidence with clinical trial data to inform best practices. FDA also expects to issue guidance soon on the use of real-world evidence to support some drug approvals and postmarketing requirements, where presumably some questions could be answered. (Also see “US FDA Wants More Examples Of Real-World Data Use” – Pink Sheet, 5 Aug, 2018.)
FDA also is funding several demonstration projects involving real-world data (Also see “Real World Evidence Benefits, Limits Explored In US FDA Demonstrations” – Pink Sheet, 29 Oct, 2017.), including an effort to replicate positive and negative clinical trials using claims databases. (Also see “Real-World Data Could Get Boost From Trial Replication Project” – Pink Sheet, 26 Apr, 2018.)
In addition, FDA Commissioner Scott Gottlieb has asked payers to help the agency access the real-world data they house, such as postmarket safety information among oncology patients. (Also see “US FDA Seeking More ‘Open Access’ To Real World Data From Payers” – Pink Sheet, 28 Jun, 2018.)
Many Useful Data Points Not In EMR
Analysts use many sources to generate and verify patient mortality, such as published obituaries and social security data. Social media also is mined for death information, along with state death registries and even credit bureaus.
Amy Abernethy, chief medical officer, chief scientific officer and senior VP, oncology, at Flatiron Health Inc., said clinics also send condolence cards to families following a death as closure for its staff, which improves death reporting.
However, even when given permission to access death information, data providers sometimes hesitate because of privacy concerns.
“I’ve been really pushing the National Academy of Medicine to think about this and try and push it forward,” Abernethy said. “I think from a policy perspective we should figure out where to concentrate our voices to be able to try and push this conversation forward.”
Other data points that could be useful real-world information also are not included in electronic medical records, such as tumor histology. Lawrence Kushi, director of scientific policy in the Kaiser Permanente Division of Research, said items such as stage of disease also are not included, even though they drive treatment.
FDA Experience Continues To Grow
FDA already has extensive experience using real-world data in drug development, often as a control arm for a clinical trial in a rare disease.
Amgen Inc.’s Blincyto (blinatumomab) and EMD Serono Inc./Pfizer Inc.’s Bavencio (avelumab), both recently approved for orphan indications using real-world evidence, are expected to eventually become blockbuster products. The postmarketing trials required as part of their accelerated approvals could validate some real-world data concepts. (Also see “Real-World Evidence At US FDA: Bavencio, Blincyto Approvals Point Way Toward Broader Use ” – Pink Sheet, 7 Aug, 2018.)
FOCR also is hoping advancements in real-world data use can open oncology clinical trials to more patients. (Also see “Real-World Evidence Could Help Relax Clinical Trial Enrollment Criteria” – Pink Sheet, 4 Jul, 2018.)