Medtronic’s defibrillator radio flaw was answered in the device clinic, not the network

The March 2019 Conexus advisory had no patch, no network fix and no reported harm. Its real response ran through cardiology: finding patients, telling them to keep monitors plugged in, controlling programmers and, later, patching devices at office visits.

Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.

Two-minute summaryThe dek and the first sentence of each section, quoted from the article

The March 2019 Conexus advisory had no patch, no network fix and no reported harm. Its real response ran through cardiology: finding patients, telling them to keep monitors plugged in, controlling programmers and, later, patching devices at office visits.

  1. The week On March 21, 2019, the Food and Drug Administration issued a safety communication about cybersecurity vulnerabilities in the wireless telemetry used by Medtronic’s implantable cardiac devices, clinic programmers and home monitors.
  2. The threat as the agencies sized it The advisory’s scoring and its narrative pointed in different directions, and the narrative mattered more.
  3. The patient list The FDA’s audience list was specific: patients with an affected Medtronic ICD or CRT-D, their caregivers, and the cardiologists, electrophysiologists, cardiac surgeons and primary care physicians treating them.
  4. The conversation The FDA’s advice to patients and providers contained a point that could easily have been lost: keep the home monitor plugged in.
  5. The equipment the hospital holds The clinic programmer was the part of the system on hospital premises, and the FDA’s guidance for it was where IT and security had a direct role.
  6. The fix delivered in the visit In March 2019, the agencies said no reprogramming or updating of the devices was required and that Medtronic was developing updates, subject to regulatory approval.
  7. Governance: joining the clinical and security halves The parts above divide the work between cardiology, IT and security.
  8. Objections Four objections deserve answers.
  9. A standing plan for implanted device advisories The Conexus advisory was specific to one manufacturer’s protocol, but its shape can recur with any connected implant.
  10. With hindsight The advisory did not end in March 2019.

The week

On March 21, 2019, the Food and Drug Administration issued a safety communication about cybersecurity vulnerabilities in the wireless telemetry used by Medtronic’s implantable cardiac devices, clinic programmers and home monitors. The same day, the Department of Homeland Security’s National Cybersecurity and Communications Integration Center published advisory ICSMA-19-080-01, “Medtronic Conexus Radio Frequency Telemetry Protocol.”

The affected devices were Medtronic implantable cardioverter defibrillators (ICDs) and cardiac resynchronization therapy defibrillators (CRT-Ds). The FDA explained that these devices provide pacing for slow heart rhythms and electrical shocks or pacing to stop dangerously fast ones, and are implanted under the skin of the upper chest with leads into the heart. Its list named 19 ICD and CRT-D entries, among them the Evera, Viva, Visia AF, Protecta, Secura, Virtuoso and Concerto lines, along with the CareLink 2090 programmer used in clinics and the MyCareLink (models 24950 and 24952) and CareLink (model 2490C) home monitors. The communication did not apply to pacemakers, cardiac resynchronization pacemakers, CareLink Express monitors or the CareLink Encore programmer.

The problem was in the Conexus protocol, which uses radio frequency to let programmers and monitors read data from an implanted device, transmit it to a clinic, display it in real time and program device settings. The FDA said the protocol “does not use encryption, authentication, or authorization,” and that the vulnerabilities, if exploited, could allow an unauthorized individual to access and potentially manipulate an implantable device, home monitor or clinic programmer. The DHS advisory assigned two identifiers: CVE-2019-6538 for improper access control, scored 9.3 under CVSS version 3, and CVE-2019-6540 for cleartext transmission of sensitive information, scored 6.5. It said successful exploitation could include the ability to read and write any valid memory location on the affected implanted device.

Both documents set limits on the risk. Exploitation required a radio device able to speak the protocol, short-range adjacent access, and the device’s radio to be active. The advisory said Conexus sessions before implant and at clinic visits had to be started by an inductive protocol, and that outside those settings the implanted device’s radio was enabled for brief periods to support scheduled transmissions and notifications. It said no known public exploits specifically targeted the vulnerabilities. The FDA said it was not aware of any reports of patient harm.

The FDA’s recommendations were operational. It told providers to keep using the programmers, noting that no programmable setting allowed a clinician to turn off the Conexus wireless capabilities; to maintain control of programmers within the facility; to use only equipment obtained directly from the manufacturer; to operate programmers within well-managed IT networks; and to remind patients to keep home monitors plugged in, because the benefits of remote monitoring outweighed the practical risk of exploitation. It said reprogramming or updating the devices was not required at that time and that prophylactic replacement was not recommended. It also asked clinicians to discuss cybersecurity risk with patients before implanting these devices. Medtronic, the advisory said, had applied additional controls for monitoring and responding to improper use of the protocol, with further mitigations to come through future updates, subject to regulatory approval.

A hospital security team reading the March 2019 advisory would have found little to do in its usual tools. There was no patch to deploy. The affected implanted devices were not on the hospital network. The home monitors sat in patients’ homes. The clinic programmers were the one component on hospital premises, and the guidance for them was to keep them under control and on well-managed networks. Everything else the agencies asked for ran through cardiology: the device clinic, the electrophysiologists, the nurses who manage remote monitoring, and the conversations they have with patients.

Our argument is that the Conexus response was a clinical workflow with a security purpose, and that health systems should plan for implanted device vulnerabilities on that basis. The security team’s role is to support the clinic, not to run the response from the network. We break the response into five parts, add the governance that joins them, answer the objections, and close with a standing plan:

  • The threat as the agencies sized it: what an attacker would need, and why the answer shaped every other step.
  • The patient list: knowing which patients carry an affected device, which is a clinical record and not an IT asset list.
  • The conversation: telling patients what to do, including the counterintuitive advice to keep monitors plugged in.
  • The equipment the hospital holds: programmers and the networks around them, the part that does belong to IT.
  • The fix delivered in the visit: how later device updates reached patients, and why scheduling became a security control.
  • Governance: how the clinical and security halves are joined.
  • Objections: where the argument may be weakest.
  • A standing plan: how the five parts become a routine.

The parts follow the order in which a health system would need to act: understand the risk, find the patients, talk to them, secure what it controls, and then deliver fixes as they arrive.

The threat as the agencies sized it

The advisory’s scoring and its narrative pointed in different directions, and the narrative mattered more. A CVSS score of 9.3 for the access control flaw signals severity. But the advisory spelled out three conditions for exploitation: an attacker needed a radio device able to transmit or receive Conexus telemetry, such as a monitor, programmer or software-defined radio; adjacent short-range access to the target; and a device in a state where its radio was active.

Reading the score

The access control flaw’s vector string, published in the advisory, recorded an adjacent attack vector, low attack complexity, no privileges or user interaction required, a changed scope, and high integrity and availability impact with no confidentiality impact. The confidentiality flaw’s vector recorded the same adjacent attack vector, an unchanged scope, and high confidentiality impact with no integrity or availability impact. Read component by component, the scores say what the narrative says: the danger of the first flaw was to the device’s function, the danger of the second was to the data, and both required the attacker to be nearby. A health system that triaged by the headline number alone would have ranked the advisory alongside internet-facing flaws that any attacker anywhere could use. Reading the components puts it where it belongs: serious, local, and dependent on physical circumstances the health system and its patients can influence.

We would ask whoever triages medical device advisories to record the attack vector and the physical conditions alongside the score, so that the clinical owner sees the same picture the security analyst does.

Distance and timing as controls

The second and third conditions are physical. An attacker had to be close to the patient, the monitor or the programmer, and had to be there when the radio was on. The advisory explained that before implant and at clinic visits, telemetry sessions required initiation by an inductive protocol, and that outside those settings the implanted device enabled its radio for brief periods for scheduled transmissions and notifications. The FDA described the same features as part of the system’s design: the protocol could be activated only by the patient’s provider at a clinic, activation times varied by patient, and an unauthorized user would need to be close to an active device, monitor or programmer.

For a health system, that sizing tells it where the risk concentrates. It is near the equipment at the moments it is talking. That means clinics during device checks, the places where programmers are stored and used, and patients’ homes during scheduled transmissions. It does not mean the hospital’s internet perimeter, its email system or its data center. A response plan that starts from the network would be looking in the wrong place.

What the impact could be

The other side of the sizing was impact. The advisory said successful exploitation could include the ability to read and write any valid memory location on the implanted device, and therefore affect its intended function. The FDA said an unauthorized person could access and potentially manipulate the device. For a device that delivers shocks to stop dangerous heart rhythms, the consequences of manipulation could be severe. We think that combination explains why the agencies published, even with no known exploit and no reported harm.

A health system’s risk assessment has to hold both halves: a narrow path to exploitation and a serious consequence if it happens. We think the right response to that combination is targeted rather than general. The organization does not need to treat every patient with an affected device as under attack. It does need to know who those patients are, make sure the equipment it controls is not the weak link, and be ready to act if the agencies’ assessment changes.

The confidentiality flaw

The second identifier, CVE-2019-6540, was about eavesdropping. Because the protocol did not encrypt, an attacker within range could listen to communications, including sensitive data. That flaw carried a lower score, 6.5, and attracted less attention. For a health system it raises a different question: whether patient health information transmitted between device, monitor and programmer could be captured. The agencies did not say any had been. But a privacy officer reviewing the advisory would want the flaw recorded in the organization’s risk analysis, alongside the safety issue, because the two may be handled by different committees.

The manufacturer’s network is part of the system

The FDA described the MyCareLink monitor as reading data from the implanted device and sending it to the patient’s physicians through the CareLink Network, over a continuous landline, cellular or Wi-Fi internet connection. The advisory concerned the radio link between device and monitor, not that network. But the description is a reminder that the system a health system relies on for remote monitoring extends from the patient’s chest, through a monitor in the patient’s home, through a manufacturer’s service, to the clinic’s screens. The hospital controls one segment of that chain, the clinic end and its programmers.

For the segments it does not control, the health system’s tools are contractual and procedural. It can ask the manufacturer how the monitoring service is secured, how advisories will be communicated, and how quickly. It can make sure the clinic knows whom to call at the manufacturer when an advisory arrives. We would add the remote monitoring service to the hospital’s list of third parties that handle patient data, with the same review any such vendor receives, because clinical value and data flows both depend on it.

The patient list

The FDA’s audience list was specific: patients with an affected Medtronic ICD or CRT-D, their caregivers, and the cardiologists, electrophysiologists, cardiac surgeons and primary care physicians treating them. To act on the communication, a health system had to know which of its patients were in that group.

A clinical register, not an asset inventory

Implanted devices are not IT assets. They do not appear in endpoint management tools or network scans. The information about which patient has which device typically lives in clinical systems: the implant record, the device clinic’s follow-up database, the manufacturer’s remote monitoring service, and the electronic health record. Pulling a list of patients with the affected models means querying those systems by manufacturer and model family.

That task belongs to the device clinic or the cardiology service line, with help from whoever manages the health record’s implant data. The security team may not have access to these systems, and should not need it. Its role is to make sure the clinic knows an advisory exists, understands the affected models, and has a way to report back how many patients are affected and how they have been contacted.

Model families and naming

The FDA’s list named device families with “all models” after each one, and separated some lines into MRI and non-MRI variants. A query that searches for one form of a name may miss another. We would ask the clinic to search by every family named in the communication, check the result against the manufacturer’s records where possible, and keep the query so that it can be rerun when the list changes. The list did change, as later updates to the advisory showed.

Patients outside the system

Some patients with an affected device will have had it implanted elsewhere, or followed elsewhere, and may come to the health system for other care. A cardiology clinic may not see them. An emergency department or primary care practice may. We think the communication’s inclusion of primary care physicians in its audience is a reminder that the patient list is not confined to the device clinic. A note in the patient’s record, visible to any clinician, can carry the advisory to wherever the patient turns up.

What the clinic needs from security

The clinic can only act on an advisory it knows about. Medical device advisories can arrive through several channels: the FDA, CISA, the manufacturer’s own security bulletins and customer letters, and industry information sharing groups. A device clinic may receive some of these and not others. We would make the security team responsible for watching all of them, and for passing each relevant one to the clinic in a form it can use: the affected models in plain product names, the agencies’ assessment of the risk, the recommended actions for providers and patients, and a named person to answer technical questions. A forwarded advisory full of vector strings and identifiers asks the clinic to do the translation; a one-page summary does it for them.

The flow should run both ways. Clinicians may hear about device problems from patients or manufacturer representatives before an advisory is published. A route for the clinic to report possible security issues to the security team, without having to judge first whether they are security issues, closes the loop.

Keeping the list current

A patient list built for one advisory goes stale. Patients have devices implanted, replaced and explanted; they move, change clinics and die. A list pulled in March 2019 would not match the population a year later. We think the useful asset is the query, not the list: a saved search in the implant and follow-up records that can be rerun for any manufacturer and model family, with a named owner who knows how to run it and how to check its results. The query should be tested from time to time, by running it for a device line and checking the count against the clinic’s own knowledge, so that it can be trusted on the day it is needed.

The same query serves other purposes: product recalls, manufacturer safety notices and research. Building it for cybersecurity advisories adds a use to a capability the clinic may already want for clinical reasons, which helps when asking for the time to build it.

The conversation

The FDA’s advice to patients and providers contained a point that could easily have been lost: keep the home monitor plugged in. The communication said the monitor must remain powered on to ensure timely transmission of alerts programmed by the physician and of automatically scheduled remote transmissions, and that the benefits of remote wireless monitoring outweighed the practical risk of an unauthorized user exploiting the vulnerabilities.

Why the advice cuts against instinct

A patient who hears that their heart device has a wireless security flaw may reasonably think the safe thing to do is turn off the wireless equipment. The FDA said the opposite. Remote monitoring carries clinical value: it lets the care team see device alerts and scheduled data without a visit. Unplugging the monitor would remove that value in exchange for a reduction in a risk the agencies judged to be small.

This is a clinical judgment, and it has to be delivered by people patients trust about their heart care. A letter from the hospital’s information security office would carry the wrong voice. A call or message from the device clinic, in language reviewed by the cardiologists, carries the right one. We think the security team’s contribution is to supply accurate facts about the advisory and to review the message for technical accuracy, not to write it.

The rest of the patient advice

The FDA’s other recommendations to patients were practical: use only monitors obtained from the health care provider or the manufacturer, keep track of the monitor, get medical help right away for symptoms such as lightheadedness, dizziness, loss of consciousness, chest pain or severe shortness of breath, and contact Medtronic with questions. The advisory added Medtronic’s recommendations: maintain physical control over home monitors, do not connect unapproved devices to them through USB ports or other physical connections, use them only in private environments such as a home, and report concerning behavior to the provider or manufacturer.

Each of these turns physical security into patient instructions. Keeping track of a monitor and using it at home are ways of keeping an attacker from being near it. A patient who understands why may follow them more closely than one who simply receives a list.

Talking about risk before implant

The FDA also asked clinicians to discuss the risk of cybersecurity vulnerabilities with patients before implanting ICDs and CRT-Ds, along with the device’s other risks and benefits. That moves cybersecurity into informed consent. We think it was a significant request, because it means the cardiologist, as well as the security team, needs a working understanding of the device’s cyber risk. Health systems may want to give their implanting physicians a short, plain summary of known advisories for each device line they use, kept current, so that the conversation is consistent across clinicians.

Recording what was said

Every patient contact about an advisory should leave a record: who was contacted, by what means, what was said, and whether the patient had questions or concerns. That record protects the patient, who may later need a clinician to know what they were told, and it protects the health system, which may need to show that it acted on the FDA’s communication. A templated note in the health record is enough. It also lets the clinic follow up with patients who could not be reached, which is where gaps are likely to appear.

Caregivers and patients who need support

The FDA’s audience included caregivers. Some patients with these devices may rely on family members or care staff to manage their home equipment, and some may have limited ability to understand technical messages. A message that works for one patient may not work for another. Clinics may want a short version for patients, a version for caregivers that explains the practical steps, and a script for staff who take follow-up calls, so that the advice is consistent whoever answers.

The equipment the hospital holds

The clinic programmer was the part of the system on hospital premises, and the FDA’s guidance for it was where IT and security had a direct role.

Physical control

The FDA told providers to maintain control of CareLink programmers within the facility at all times according to hospital IT policies. Medtronic’s recommendations, quoted in the advisory, added that programmers should be used to connect and interact with implanted devices only in physically controlled hospital and clinical environments. A programmer is a device that can write to an implanted defibrillator. Its location and custody are a security matter.

In practice that means an asset record for each programmer, a known storage location, a sign-out process if programmers move between clinics or wards, and a check that none has left the building. These are ordinary controls for valuable clinical equipment. The advisory gave them a security reason.

Network placement

The FDA told providers to operate programmers within well-managed IT networks and to consult their IT departments on network security, pointing to the NIST Cybersecurity Framework for recommended actions. The advisory’s recommendations from NCCIC were general: restrict system access to authorized personnel, follow least privilege, apply defense-in-depth strategies, and disable unnecessary accounts and services. For a health system, that points to placing programmers on a segmented network for clinical devices, limiting what they can reach and what can reach them, and monitoring that segment.

The FDA communication also referred to an October 2018 safety communication about a software update to address vulnerabilities in Medtronic’s cardiac device programmers. A hospital that had applied that update had already dealt with one programmer issue. The March 2019 advisory was a reminder that programmers needed their own patch and configuration record, maintained with the manufacturer, separate from the hospital’s general computer patching.

Programmers at the end of their life

A programmer that leaves service still matters. It is a device built to communicate with implanted defibrillators, and the advisory named the programmer itself as one of the affected products. A health system retiring or replacing programmers should return them through the manufacturer or dispose of them through a controlled process, record the disposal, and make sure none is sold or given away through general surplus channels. The same applies to home monitors returned by patients. The asset record that tracks a programmer in service should also show how it left.

Supply integrity

Both the FDA and Medtronic told users to obtain monitors, programmers and implantable devices directly from the manufacturer or the health care provider to ensure the integrity of the system. For a hospital, that is a procurement control. Programmers and monitors should come through the manufacturer’s channel, not from secondary markets or informal transfers between facilities. The supply chain record for each programmer is part of its security record.

The fix delivered in the visit

In March 2019, the agencies said no reprogramming or updating of the devices was required and that Medtronic was developing updates, subject to regulatory approval. Later updates to the advisory showed how fixes would reach patients, and the answer put security on the clinic schedule.

Patches installed during office visits

The later version of the advisory, Update C, last revised April 8, 2021, said Medtronic had developed mitigating patches for a subset of the affected implanted device models, and that the patches were installed during regular office visits. It listed the models for which patches were available, including the Evera, Viva, Visia AF, Brava, Mirro MRI, Primo MRI and Protecta lines, and noted that for some CRT-D models the patch was available in the United States only. It said patches for additional models were being developed.

For a health system, that meant a security update reached each patient through an appointment. The patch had to be applied by clinic staff, with a programmer, during a visit. A patient who missed follow-up visits would not receive it. The device clinic’s scheduling, recall and no-show processes became the delivery mechanism for a security fix.

Tracking the fix per patient

We would treat each patch release as a clinical task with a register: the patients whose device model has a patch available, the date of their next scheduled visit, whether the patch was applied, and the reason if it was not. The clinic may already track device checks this way. Adding the patch status is a small change. Without it, the organization cannot say what share of its affected patients carry the mitigated version, and cannot answer a regulator, a patient or its own board if asked.

Models without a patch

Update C’s list did not cover every affected model. For patients with models not yet patched, the March 2019 measures remained the protection: physical control of monitors and programmers, monitoring by the manufacturer, and the design limits on when the radio was active. A health system’s register should show those patients separately, so that if new information raises the risk, the clinic knows exactly whom to call.

Patients who cannot attend

A fix that depends on a visit reaches patients who come in. Some do not: patients who are frail, who live far from the clinic, who have moved, or who have stopped attending follow-up. For them, the patch status register will show a gap that the normal recall process may not close. We would ask the clinic to review those patients as a group, decide whether the patch warrants an extra outreach effort, and record the decision. In some cases the right answer may be that the patient’s next routine visit is soon enough, given the agencies’ assessment of the risk. In others, the clinic may decide to make a special effort. Either way, the decision should be made by a clinician and written down, rather than left to the scheduling system.

Governance: joining the clinical and security halves

The parts above divide the work between cardiology, IT and security. That division only works if someone joins the halves together. We would make that explicit, with three elements.

A named risk owner

Each implanted device advisory should have a named clinical owner, such as the medical director of the device clinic, and a named security owner. The clinical owner decides how patients are contacted and how fixes are delivered. The security owner assesses the advisory, secures the programmers and reports status. Both report to one executive who can settle disagreements, whether that is the chief medical officer, the chief nursing officer or a patient safety committee chair.

One status report

Leadership should receive one report per advisory, not two. It would show the advisory and the agencies’ risk assessment, the number of affected patients identified, the number contacted, the programmer status, and the patch status where patches exist. A single report keeps the clinical and technical work in view together and makes gaps visible: a patient list with no contact record, or a patch released with no delivery plan.

The right committees

Implanted device advisories touch several committees that may not meet together: patient safety, medical device or clinical engineering, information security, and privacy. We would route each advisory to all of them with a short summary, and ask one to lead. In our reading, the March 2019 advisory was mainly a patient safety matter with a security cause, which suggests patient safety as the lead. A different advisory, centered on data exposure, might be led by privacy. The choice matters less than making it.

Objections

Four objections deserve answers.

The risk was too small to justify this effort

The first objection is that an adjacent-access radio attack with no known public exploit and no reported harm does not warrant a patient register, a communication campaign and a patch-tracking process. We have some sympathy with this. The agencies themselves told patients to keep using their devices and monitors. But the work we describe is modest, and much of it reuses processes many device clinics already run: patient lists, recall and follow-up scheduling. The cost of having the list ready is small. The cost of building it under pressure, if the risk assessment changed, would be larger.

Security teams should own security issues

The second objection is that putting cardiology in charge of a cybersecurity response blurs accountability. If the device clinic runs the patient side, who is accountable for the security outcome? Our answer is that accountability can be shared by function. The security team owns the advisory intake, the technical assessment, the programmer network and the reporting to leadership. The clinic owns patient identification, communication and patch delivery. Both report to the same risk owner, who may be a chief medical officer, a chief nursing officer or a patient safety committee. The objection would be right if no one joined the two halves together.

Patients may be frightened

The third objection is that telling patients about a security flaw in a heart device may cause anxiety out of proportion to the risk, and lead some to unplug monitors or skip care. That risk is real, and it is why the message should come from the clinicians and lead with what the patient should do. The FDA’s own framing did this: continue to use the device as intended, keep the monitor plugged in, and seek care for symptoms. A health system that stays silent does not prevent anxiety. Patients may hear about an advisory from news coverage, and then have only that account to go on.

The manufacturer should carry the burden

The fourth objection is that the device maker designed the protocol and should bear the cost of fixing it. That is a fair point about responsibility, and later law moved some of the burden to manufacturers, as the hindsight below describes. But the patients belong to the health system’s care, and the fixes reach them through its clinics. Whatever the allocation of cost, the health system holds the patient relationship and the appointments through which the work gets done.

A standing plan for implanted device advisories

The Conexus advisory was specific to one manufacturer’s protocol, but its shape can recur with any connected implant. We think a health system would benefit from a standing plan with five elements that match the five parts above.

For the threat, a short assessment template that asks the questions the March 2019 advisory answered: what access an attacker needs, when the device is exposed, what an attacker could do, and whether exploitation or harm has been reported. For the patient list, a saved query in the implant and follow-up records for each manufacturer and model family the health system implants or follows. For the conversation, a template message, reviewed by cardiology, that leads with what patients should keep doing. For the equipment, an asset record for each programmer and monitor the hospital holds, with location, custody, network placement and update status. And for fixes, a patch status field in the device clinic’s follow-up record.

Testing the plan

A plan that is used once every few years will decay unless it is exercised. We would run a short tabletop exercise each year with the device clinic, security, clinical engineering and patient safety, using a past advisory such as this one. The exercise would check that the patient query still runs and returns a believable count, that the template message can be approved quickly, that the programmer inventory is accurate, and that the patch status field exists and is being used. An hour spent on that once a year is a small price for confidence that the plan will work when a new advisory arrives.

The exercise is also a chance to update the plan for new device lines. When the health system starts implanting a new model or following patients from a new manufacturer, the query, the message templates and the programmer inventory should be extended to cover it. The exercise gives that a regular date.

The security team’s role in that plan is real but supporting. It watches for advisories, assesses them, keeps the programmer network sound and reports the overall status. The clinic does the work that reaches patients. The March 2019 advisory showed that, for this class of device, protecting patients from a cybersecurity flaw is done in the clinic, one visit at a time.

With hindsight

The advisory did not end in March 2019. CISA’s later version, Update C, last revised April 8, 2021, followed an Update B published June 4, 2020. Update C described mitigating patches for a subset of the affected implanted device models, installed during regular office visits, with some available in the United States only, and said patches for further models were being developed. Its list of affected products added the Brava CRT-D, which neither the original advisory nor the FDA’s March 2019 list had named, and listed the Mirro MRI ICD by that name next to the Mirro ICD the original advisory had listed. It also stated that the vulnerabilities were not exploitable remotely.

The legal footing for device cybersecurity changed later. The FDA’s cybersecurity FAQ says Congress granted the agency explicit cybersecurity authority in 2022 by adding section 524B to the Federal Food, Drug, and Cosmetic Act, effective March 29, 2023. Under section 524B, the sponsor of a premarket submission for a “cyber device” must include a plan to monitor, identify and address postmarket cybersecurity vulnerabilities and exploits, including coordinated vulnerability disclosure; must design, develop and maintain processes to provide a reasonable assurance that the device and related systems are cybersecure and make postmarket updates and patches available; and must provide a software bill of materials. The FAQ defines a cyber device as one that includes software, has the ability to connect to the internet, and has technological characteristics that could be vulnerable to cybersecurity threats. The requirements do not apply to submissions made before March 29, 2023, but apply to a new submission when a previously authorized device changes in a way that requires premarket review. The FAQ says that from October 1, 2023, the FDA expected sponsors of cyber devices to be ready to include the required information.

The FDA’s guidance followed. Its cybersecurity page lists a final premarket guidance issued September 27, 2023, a revised version issued June 27, 2025 that added a section on section 524B and cyber devices, and a further final version issued February 2, 2026 to align with the amended quality management system regulation. Under November 15, 2023, the same page lists a report the FDA contracted MITRE to develop on next steps for managing legacy medical device cybersecurity risks. In September 2022, the FDA alerted users to a potential cybersecurity risk associated with the communication protocol of Medtronic’s MiniMed 600 series insulin pump system.

The FDA’s list of cybersecurity safety communications carries a standing note: unless otherwise noted in the description, the FDA is not aware of any patient injuries or deaths associated with those alerts, though left unpatched or otherwise mitigated the vulnerabilities may potentially lead to patient harm.

What held up is the division of labor. The 2022 law, effective in 2023, placed duties on manufacturers of cyber devices to plan for vulnerabilities, make updates and patches available and provide a software bill of materials in their premarket submissions. The law does not itself deliver those updates into patients. For the Conexus devices, Update C described patches installed during regular office visits. What we would add today is that, for devices covered by section 524B, health systems have something to ask for at purchase: the manufacturer’s vulnerability plan and update process for each device line, which tells the device clinic in advance how future fixes will reach its patients.

Spotted an error? Tell us and we will check it and correct the article where needed.

  • Medtronic
  • medical device security
  • implantable devices
  • FDA
  • section 524B
  • patient communication