What the FDA Teaches Us About Software Changes: The Nihon Kohden Warning Letter

On June 12, 2026, the FDA issued a Warning Letter to Nihon Kohden Digital Health Solutions, the company that develops and markets Next Generation NetKonnect (NGNK), physiological monitoring software that detects cardiac arrhythmias, including ST segment measurement.

At first glance, this sounds like one particular company that made a mistake. But anyone who reads the letter carefully will see a pattern that keeps repeating. The letter reveals two completely different failures: the first relates to a decision about a code change, and the second to how the company described its product to customers.

Both came from the same place: a regulatory assessment that was too far removed from reality.

First change: a silence button that became a safety problem

The NGNK software received clearance under K220989 in July 2022. In October of that year, the company added a Silence Alarms function to the software, with the option to silence alarms. In the company's regulatory file, the regulatory manager concluded that the change would not significantly affect safety and documented the decision through a Letter to File (LTF) only, without submitting a new 510(k).

The FDA disagrees with this decision.

According to the letter, adding a silence function is a change to a control mechanism, because it changes the way the user controls the device's alarm behavior. For question B2 in the regulatory assessment, "Is the change to a control mechanism, operating principle or energy type?", the answer chosen was "No". In the FDA's view, the correct answer was "Yes", and that should have led to a new 510(k) submission.

Why does this matter so much?

Because a physiological monitoring system that lets users silence alarms, without that change having gone through a proper regulatory assessment, can mean that a critical alarm is not heard at the moment it is needed most.

The FDA says this explicitly: this change could lead to alarms for life-threatening conditions that require immediate response going unnoticed, and to real harm to patients.

Second change: "device-agnostic"

The second problem is different in nature. The NGNK labeling added the term "device-agnostic compatibility", meaning the ability to support input from other manufacturers' devices, not only from Nihon Kohden devices.

What looks like a natural marketing advantage is actually a change in intended use.

The original clearance (K220989) covered support for input from Nihon Kohden devices only. Once the company stated that the device also supports input from other devices, devices that had not been validated for compatibility with NGNK, a new intended use was created that requires an additional 510(k) under 21CFR 807.81(a)(3)(ii).

The risk the FDA points to here is unverified interoperability: a sensor that has not been validated with the software may produce incorrect measurements, cause signal loss, and ultimately lead to delayed or incorrect diagnosis.

Together, the two problems lead the FDA to conclude that NGNK is at the same time adulterated, because it has no valid premarket approval, and misbranded, because no premarket notification was submitted for bringing the modified device to market.

A related case: WHOOP and the limits of wellness

A parallel case that sheds light on the FDA's logic is the Warning Letter sent to WHOOP in July 2025. The FDA determined that the Blood Pressure Insights (BPI) feature the company marketed had not received clearance and was a medical device in every sense, even though the company argued it was only a wellness feature.

WHOOP held that the product was not a medical device, partly because it had added explicit statements to the user interface saying that BPI is not intended for diagnosis. The FDA did not accept this argument. As long as the labeling points to the legal definition of a medical device, that is, diagnosing conditions and/or monitoring health, disclaimers are not enough to change the classification.

The Warning Letter was closed in June 2026, after WHOOP changed the feature in line with the updated General Wellness guidance the FDA published in January 2026. The FDA relaxed its approach somewhat for wearables with non-invasive measurements in a wellness setting, and WHOOP adapted accordingly. But the point made in 2025 still stands: intended use is determined by the labeling and by what the device does, not by what the company prefers to call it.

What the numbers say

According to an analysis by Emergo by UL of the FDA's Warning Letters for 2025, out of 695 letters the FDA issued, about 54 concerned medical devices. Of these:

  • 7 dealt with devices without clearance or approval

  • Another 7 dealt with a clearance that does not cover the current device or labeling

  • 7 dealt with design control and change control deficiencies, including risk management

One trend stands out: over the past two years, the FDA has stepped up enforcement around questions of intended use and changes to control mechanisms, and not only around classic quality system failures such as corrective actions and supplier controls. This does not mean quality system topics matter less. It means enforcement on changes made without a 510(k) submission has grown.

What this means in practice

Anyone responsible for change control in medical device software, whether they work at the company or advise it, should come away from this case with a few clear questions:

  1. Does your change management process force you to ask the right question? Question B2 in the change assessment for NGNK was not analyzed correctly. This is not only the regulatory manager's mistake, but a process deficiency. A good software change management process must ask explicitly: Does the change alter the way the user controls the device? This applies even when the change looks minor.

  2. What does "device-agnostic" mean in validation terms? Any expansion of interoperability, or support for input from sources not covered by the clearance, is a change in intended use. Without specific compatibility validation, this is not a change that can be documented in an LTF. The FDA expects to see a new 510(k) that defines and justifies the compatibility.

  3. Is the labeling aligned with the intended use in the clearance? The WHOOP case shows that any change in labeling, including marketing descriptions, what is written on the website, and even wording in the user interface, can determine intended use. A routine review of marketing materials as part of the release process is not a luxury.

  4. When is a Letter to File no longer enough? The FDA refers to 21CFR 807.81(a)(3) as the basis: a change that significantly affects safety and effectiveness requires a 510(k). Control mechanism and intended use are two categories that will usually require a new submission, even when the change does not feel "big enough" from the inside.

When a Letter to File is not enough

The Nihon Kohden case is especially relevant for any company that markets medical software. An alarm silence button and a marketing sentence about device-agnostic connectivity are not minor technical details. They are changes that alter what the FDA cleared. When the regulatory manager does not recognize this in time, the risk is not only regulatory exposure but also real risk to patients.

The Letter to File is an important tool, but it is only enough when the analysis behind it is rigorous enough.

Whenever there is doubt about a control mechanism, intended use, or compatibility validation, the safe path is to ask the right questions rather than make assumptions.