Deaths associated with some transcatheter valve-repair procedures may be underreported in a US Food and Drug Administration (FDA) adverse events database, leaving a misleading picture of the number of associated fatalities, a new report suggests.
It found that 17.5% of deaths associated with the SAPIEN 3 (Edwards Lifesciences) transcatheter valve and 24.7% of those associated with MitraClip (Abbott Vascular) were misclassified as “injury” or “malfunction” events in the FDA’s Manufacturer and User Facility Device Experience (MAUDE) database.
“We found that a significant number of deaths associated with these high-risk cardiac devices were not correctly classified,” senior author Rita F. Redberg, MD, MSc, told theheart.org | Medscape Cardiology.
“Therefore, clinicians, patients, or anyone searching MAUDE — the primary source for adverse event data — to determine how many deaths were reported associated with these devices would get an erroneously low number for deaths reported to MAUDE, which is already only a small fraction of all adverse events, as most are not reported to MAUDE,” said Redberg, from the University of California, San Francisco.
Redberg is also editor of JAMA Internal Medicine, in which the study was published October 7.
“Cumbersome” Process, Low Reporting
Postmarket surveillance data re increasingly central in informing clinicians and the public about the safety or potential adverse events of medical devices, as the FDA “moves to hasten” their approval, the authors write.
That leaves “less time for premarketing clinical studies” and heavier reliance on postmarketing data as the “principal way adverse events and risks become apparent.”
However, “the process of reporting adverse events is cumbersome and reporting rates are low,” they note. And, although device-user facilities are required to report adverse events to the FDA, reporting by physicians is voluntary.
Summaries submitted to the FDA by the Transcatheter Valve Therapy (TVT) registry “may omit redacted information and obscure important data,” the authors say. The registry does not make its data publicly available, they add, thereby limiting its value.
The publicly accessible MAUDE database allows adverse events to be categorized as “malfunction,”http://www.medscape.com/”injury,” or “death.”
“We are interested in patient safety and ensuring optimal use of high-risk medical devices,” Redberg said. “Such use requires accurate data on risks and benefits, and in this study, we were looking at the risks.”
To investigate the MAUDE classification system, the researchers used the software Device Events, a web-based tool that searches medical device postmarketing surveillance data and “collates adverse event reports from MAUDE in a more accessible format.”
The researchers analyzed adverse event reports on the Sapien 3 and MitraClip from the time of their FDA approval, in 2013 and 2015, respectively, through 2018.
They turned to the “critical-events thesaurus” within Device Events to identify terms synonymous with death that may have been used in event reports — words such as “expired,”http://www.medscape.com/”hospice,”http://www.medscape.com/”comfort care,”http://www.medscape.com/”passed away,”http://www.medscape.com/”died,”http://www.medscape.com/”autopsy/autopsied,” and “comfort.”
The researchers looked at whether deaths were categorized as “death” or instead were noted in entries classified as “injury” or “malfunction.”
For Sapien 3 devices, they found 9320 injury and malfunction reports (and 1021 “deaths”). However, 217 (2.3%) of those reports also stated the patients had died during or after implantation of the device.
“In addition to directly using the word died, the most commonly used terms in these reports to describe the death of a patient were expired, passed away, and autopsy,” the authors note. “Thus, misclassified reports made up 217 of 1238 (17.5%) total patient deaths.”
For MitraClip devices, they found 5323 injury and malfunction reports (and 295 “deaths”). However, 97 (1.8%) of the injury and malfunction reports also said the patient died.
“Terms used in injury and malfunction reports to describe patient deaths included expired, hospice, and passed away,” they report. “Thus, 97 of 392 (24.7%) patient deaths were misclassified as injury or malfunction events.”
The findings are “worrisome,” Redberg said. “The deaths are often related to a device-related injury or malfunction, although that cannot be assumed. But if a death occurred, the report should be in the ‘death’ category,” she said.
A spokesperson for the FDA told theheart.org | Medscape Cardiology that the agency “takes seriously its responsibility to monitor device safety and performance, take action when needed, and help ensure patients and providers have the information they need to make informed healthcare decisions.”
The spokesperson noted that the medical device reports, available to the public through the MAUDE database, “are just one of the tools FDA uses to monitor device performance, detect potential device-related safety concerns or signals, and contribute to the benefit-risk assessment of these products.”
This passive reporting system has limitations, potentially leading to “miscategorization of events,” the spokesperson acknowledged. But they are also “closely monitored by FDA analysts who do not rely solely on the event categorization by the reporters.”
Additionally, the FDA has “advocated for development of a modernized, active surveillance system that utilizes real-world data for the future of device signal detection.”
Efforts are also underway to improve the “accuracy and clarity” of adverse event reporting, including increased focus on manufacturer compliance with reporting requirement, and making medical device reports more transparent to the public, the spokesperson added.
A statement to theheart.org | Medscape Cardiology from Abbott Vascular said that the company “ensures all information is appropriately reported to the MAUDE database inclusive of all patient outcomes.”
The MAUDE database, the statement adds, “is not a comprehensive representation of product performance.”
The company statement notes that the current study “uses MAUDE data to connect MitraClip to patient deaths of unconfirmed causes, when the product itself was unrelated to their death.”
Edwards Lifesciences, in a statement provided to theheart.org | Medscape Cardiology, said that “as a condition of FDA approval of transcatheter aortic valves in the US, as well as reimbursement coverage, every patient treated with a Sapien 3 valve in the US has data and outcomes reported by the hospitals directly to the…TVT Registry database,” which is run by the American College of Cardiology (ACC) and the Society of Thoracic Surgeons (STS).
That independent registry itself “tracks and monitors all post-approval TAVR patient follow-up, with regular reports to the FDA.”
Sorting Out Attributions
James McCabe, MD, stated for theheart.org | Medscape Cardiology that he is not “necessarily surprised by the presented results, which likely reflect how challenging it can be to sort out attributions; for example, did someone die with the prosthesis or from the prosthesis?”
Nevertheless, “it’s worth remembering that the devices in question are not only assessed in multiple postmarket registries (TVT and MAUDE), but also have been studied in multiple rigorous randomized clinical trials,” he continued.
McCabe, who directs the cardiac catheterization laboratory at University of Washington Medical Center, Seattle, was not involved in the current study.
“The randomized data for both of the devices in question, particularly the [transcatheter aortic-valve replacement] technology, is perhaps unparalleled and informs most of our understanding of the risks and benefits of the technologies.”
He noted that although MAUDE reporting is optional for physicians, it is mandatory for device manufacturers, which the authors “neglect to mention.”
Moreover, “TVT reporting is mandated, and reporting hospitals pay the ACC/STS to do so.”
Postmarketing data are “an enormous unfunded mandate for hospitals and physicians,” with reporting costs covering dues, information technology infrastructure, “and salary support for dedicated staff,” McCabe pointed out.
“Postmarket reporting is an important endeavor, but the lack of resources to support such programs may limit data integrity,” he said.
“Incentivizing hospital systems to invest in data-collection systems is certainly valuable, but the aggregate ongoing costs to the hospitals can be quite substantial and are often underappreciated,” said McCabe.
Redberg suggested that the FDA, “due to its budget limitations and priorities, does not have sufficient staff devoted to reviewing the texts of adverse events reports.”
“We think this should be a high priority for the agency and for appropriations, as it is central to the mission of protecting public health,” she emphasized.
No source of study funding is listed. Redberg is editor of JAMA Internal Medicine but reports she was not involved in any of the decisions regarding review of the manuscript or its acceptance. Tomes is the chief executive officer of Device Events. The other authors disclosed no relevant financial relationships. McCabe reports consulting for Edwards Lifesciences.
JAMA Intern Med . Published online October 7, 2019. Research letter