Anatomy of a medical device recall: How defective products can slip through an outdated system
This is the first in a series of stories on medical device recalls. Future stories will explore a seminal Supreme Court decision and other product safety topics.
Medical device recalls require speed to ensure patient safety.
But the system used to pull defective products with potential risk of serious injury or death is clunky and uncoordinated, and results at times in recalled devices being used by or implanted in patients.
Reliance on paper communications and mail services to deliver recall notifications, limited use of device-specific identifiers and a lack of data reporting standardization results in a system that can take weeks for hospitals to learn about a manufacturer’s recall and months for patients to discover they’re using or have been implanted with a defective product.
The system also lacks uniformity, with crucial steps varying by the company issuing a recall, the hospital addressing a recall or the specific product at issue.
Nancy LeMaster, a consultant specializing in the healthcare supply chain, said while consumers can check on recalls for products like peanut butter directly on their smartphones as they shop, medical device recalls are burdened by decades-old communication methods.
“People would be astounded to know that it can take the FDA 90 days to even categorize a recall, that all of that information is done on paper, that it is transcribed by the FDA, by the providers, by the distributors, by third-party recall services, [and] the level of error involved,” LeMaster said. “It truly is like we are operating in about the 1950s.”
The FDA can receive over one thousand recalls every year, labeled as either a Class I, II, or III event. Class I is the most serious, suggesting “a reasonable probability that the use of or exposure to a violative product will cause serious adverse health consequences or death.”
Adding potential confusion, recall classifications are ordered opposite of device classifications: Class I devices pose the least risk to patients, followed by Class II and Class III devices.
The FDA has held two public meetings in the last 11 months to hear from patients and experts and find ways to improve the system. The meetings follow several high-profile safety incidents with devices, such as Allergan’s recall of breast implant products that caused cancer in hundreds of women, hundreds of deaths and thousands of injuries related to the use of surgical staplers, which led to the uncovering of the FDA’s shadow medical device reporting database, and most recently Philips’ recall of millions of sleep apnea and ventilator machines.
Reasons for recalls vary greatly, ranging from life-threatening malfunctions of pacemakers to simple labeling changes for non-implantable products. And the action does not always mean products must be pulled; sometimes a recall means a product needs to be checked or fixed.
The FDA can technically require a device maker to recall a product — but it rarely does, leaving the initiative to the manufacturer, which most often does so voluntarily.
An agency official speaking on background said that FDA-forced recalls are extremely rare and could likely be counted on one hand, adding that voluntary recalls are easier and faster than mandating products be pulled. Some critics, including patient groups, call that approach flawed and too deferential to industry, leaving patients unnecessarily at risk.
FDA recall classification definitions
Recall Classification | Definition | Number of 2020 recalls |
Class I | A situation in which there is a reasonable probability that the use of or exposure to a violative product will cause serious adverse health consequences or death. | 41 |
Class II | A situation in which use of or exposure to a violative product may cause temporary or medically reversible adverse health consequences or where the probability of serious adverse health consequences is remote. | 994 |
Class III | A situation in which use of or exposure to a violative product is not likely to cause adverse health consequences. | 43 |
SOURCE: U.S. Food and Drug Administration
“Unless you’re hands-on dealing with recalls — even at a hospital — a lot of people don’t realize just how broken the process is,” said Guillermo Ramas, CEO and founder of NotiSphere, a company that works with providers and suppliers to help process recalls.
Wasted time
Medical device recalls typically begin with safety or manufacturing issues identified by the device maker, physicians or patients. Such issues can be flagged to companies, providers or the FDA itself through the agency’s Manufacturer and User Facility Device Experience (MAUDE) database.
MAUDE, however, has several complications, such as incomplete or inaccurate reports, duplicated reports or reports mislabeled as “injury” when it should be marked as “death.” The FDA even acknowledges some of the problems with the database.
Reports of patient safety risks, device malfunctions or manufacturing problems are measured against risk thresholds, which companies create before a product is released. Manufacturers use these as road maps to determine whether issues identified will require an investigation and eventual recall of the product, according to an industry spokesperson.
The risk thresholds vary by device. For example, safety risks might be more serious for permanent implanted devices compared to a non-implantable device, or patient deaths linked to a product might require an immediate recall rather than the company conducting a longer investigation if patient impact is not as serious.
Michael Heyl, a partner with the law firm Hogan Lovells, said manufacturers will investigate once they receive reports of safety issues, looking at factors such as the severity of impact on patients, how many affected devices are in the field or the rate of occurrence of a malfunction.
These early steps can lack standardization and vary by manufacturer or even by product within a manufacturer, which some experts say complicates the system. However, Heyl contends the process is similar across the industry with some variation by company.
“The general process should be consistent,” Heyl said. “The numbers and where thresholds are, that’s where you’ll see some variability.”
As reports of malfunctioning devices or patient injuries or deaths come in, manufactures are required to report them to the FDA through MAUDE, along with importers and device user facilities.
If the adverse event reports add up or are serious enough to require immediate action, such where there are patient deaths, medical device manufacturers will send letters to healthcare providers that a recall has been initiated. And for a process that requires speed, wasted time is a consistent critique.
Manufacturers communicate recalls in a predominantly paper-based system, sending out notifications to providers and distributors through the mail or FedEx.
“The majority of people believe they will be promptly notified by their healthcare provider if the device in their body needs to be removed. This is not always the case.”
Rich Kucera
CEO, Symmetric Health Solutions
Joan Meléndez, CEO of Xcelrate UDI, which works with providers to identify and track recalls, said notifications often go to the wrong people in hospitals, sometimes taking weeks or months to get those actually responsible for pulling products from inventories.
In practice, devices have been implanted or used on patients after a recall has been triggered because of the time it takes for notifications to make it from the manufacturer to the appropriate people inside a healthcare facility, Meléndez and Ramas said.
“By the time it trickles down to the actual operating room, it could be months, so recalled items are being used that whole entire time … there’s a lot of failure,” Meléndez said.
Manufacturers have 10 days to deliver the first report to the FDA once a recall has been issued, and then the agency reviews information collected by the company and classifies the recall. If the FDA labels the recall as Class I or determines that a device poses a big enough threat to patients, the agency will issue a public recall notice or safety alert. However, these notifications can come several weeks after a manufacturer has sent the original recall letter to providers.
This delay in notification to classification to publication has led to patients saying they can be the last to know about a recall even though they are the most at risk; sometimes patients find out via news reports or social media.
“The majority of people believe they will be promptly notified by their healthcare provider if the device in their body needs to be removed. This is not always the case,” Rich Kucera, CEO of Symmetric Health Solutions, a software company for healthcare supply chain data, said during an Oct. 6 FDA committee meeting.
An example of this fragmented timetable can be seen in Philips’ recent recall of millions of sleep apnea machines and ventilators. The company recalled the machines due to foam used to dampen noise breaking down and possibly being inhaled or ingested by users.
The defect, which affected devices manufactured between April 2007 and April 2021, was first made public by the company this April as part of an earnings release. The company did not initiate a recall until June, and patients, some of which are suing the company, have said they were never contacted by physicians or Philips.
Timeline of Philips sleep apnea, ventilator device recall
-
April 26
Philips notifies the public when reporting Q1 earnings that there are “possible risks to users ” in certain sleep and respiratory care machines. The company takes a €250 million provision to cover costs.
-
June 14
Philips initiates a recall of 3 to 4 million machines, warning users they can be exposed to “possible toxic and carcinogenic effects” when the sound abatement foam breaks down and is inhaled or ingested.
-
June 30
FDA releases a safety communication for the Philips recall.
-
July 22
FDA officially labels the recall as a Class I event and publishes a public notification.
Erin Keith, associate director of compliance & quality with the Office of Product Evaluation & Quality in the FDA’s Center for Devices and Radiological Health, said during the recent committee meeting the responsibility to contact patients falls to device makers’ customers, meaning hospitals, providers, retailers and distributors.
However, providers as a point of contact is not always the most effective as patients may go to a primary care physician if there are health issues rather than the hospital or center where their operation took place.
Hospitals and providers also may not contact patients, instead waiting for them to come in with symptoms, according to Meléndez.
“There’s no clear cut accountability for who’s actually responsible for notifying a patient that they may have a recalled implant [or] that they may have been touched by a recalled device,” LeMaster said.
If that contact is not made, patients can get lost in a series of databases, recall notices, press releases and safety alerts trying to figure out if they are or are not at risk.
“The FDA actually needs to step up and become that trusted source. Because I gotta be honest here, nobody trusts the FDA anymore,” Rachel Brummert, communications lead for the healthcare group American Society of Pharmacovigilance, said at the October FDA meeting on the topic.
Brummert called on the FDA to take the lead role in notifying patients.
Connecting devices to patients
A key step is tracking down affected patients.
When Medtronic pulled its HeartWare ventricular assist device from the market in June, approximately 4,000 people had the defective product still implanted. Patients and doctors were notified by Medtronic in the announcement that the benefits of the device might outweigh the risk of removing the device even though the system had raked up numerous Class I recalls. Medtronic did coordinate with rival Abbott Laboratories, which makes a similar device, to ensure that patients have an option if the HeartWare implant needs to be taken out.
Philips had between 3 to 4 million machines to track down and set up an online registry for patients to register their machines for possible repair or replacement. Similar to Medtronic, the company said certain users should consult with their physicians before stopping use.
While manufactures can sometimes identify the majority of devices in a recall quickly, either by having them sent back or being notified from providers that they’ve been destroyed, identifying and tracking down 100% of the devices can take months and require routine communication with providers or distributors.
Identifying every device can also be challenging for manufacturers because providers can be unresponsive to notices, according to Heyl.
“It gets hard … You’ve got a large volume product and there are some people who refuse to send [a product] back, refuse to not use it or just refuse to reply,” Heyl said. “The expectation for industry is that you will just show your due diligence as to what you did to get the word out to those customers.”
The multiple communications can cause even more confusion, making it look like a separate recall was initiated.
Similar issues that hamper recall communications can make locating the correct products and patients difficult.
One hurdle is the lack of specific device identification. Recalls are typically reported in lot numbers, which can be defined differently by manufacturer. According to an industry spokesperson, a lot number may refer to an 8-hour manufacturing window, and how many individual devices manufactured during this window can vary by the product.
The lot numbers included in recall notifications can be in the tens of thousands, making it hard for workers in charge of pulling inventory to navigate, according to Meléndez.
Use of unique device identifiers to quickly identify the specific device recalled could help. UDIs are numeric or alphanumeric codes that generally contain two parts, the device identifier (DI) and the production identifier (PI).
The FDA issued a final rule in 2013 requiring UDIs to be part of product labeling, with a phased-in implementation. Currently, Class II and III medical devices and all implantable devices are required to be labeled with a UDI.
However, the code is not required to be used in recalls.
Terrie Reed, director of partnerships for Symmetric Health Solutions, said UDIs make it easier to connect specific devices to patients that products are used on or implanted into.
A device’s UDI can be scanned when taken out of inventory for use and logged in a patient’s electronic health record. When a product is then recalled, that UDI can be traced quickly from the recall to the patient.
“When that device, especially an implant, gets recalled — if you have that same device identifier information, that same UDI, then it’s really easy to match for a hospital,” said Reed, who was with the FDA for nearly 15 years, working on projects like the unique device identification system. “If the FDA is not using UDI, or structured data in their recalls, it makes that whole process more difficult.”
But UDIs are only used in a fraction of recalls, according to a report from the Learning UDI Community, a part of the supply chain arm of the American Hospital Association. The report contends that manufacturers and hospitals are not using UDIs because the FDA does not require them to be part of recalls.
Another more modern improvement to the recall process is the standardization of data format and reporting, according to Reed. What data is being reported, such as UDIs, or the format data is presented in, such as consistent date entry, varies by manufacturers. Information can also be sent in a PDF or email, which requires transcription rather than downloading structured data.
The lack of data standardization slows the system down and compromises patient safety, Reed said.
What’s next?
While the FDA is exploring requests from the two public meetings, an agency official was unable to say if there will be an overhaul of the process.
From the industry point of view, Heyl argued that the process needs variation and enforcement like more mandated deadlines or fines would not work because companies need time to investigate and identify the causes of safety issues.
Any change would need the coordination of the key players — device manufacturers, regulators, providers and distributors — which may be hard to accomplish. An industry spokesperson said they were hesitant to back a more standard reporting system because past changes to the process had numerous issues and costs typically fell on manufacturers.
An overhaul may not be needed right away as smaller changes can still have a big impact on patient safety, according to Reed.
“I recognize this is not a simple problem to solve,” Reed said. “But that doesn’t mean we can’t try to fix at least little pieces of it.”