How can a manufacturer verify that a field correction actually happened?

Availability, implementation and proof are different things

An organization can issue a notice, make an update available and schedule service without yet correcting the affected product. Each step is necessary in many campaigns, but none is equivalent to verified completion. A management dashboard that combines them into a single “actioned” category can conceal the work that remains.

My proposed principle is simple: define completion in observable terms before counting it, and keep uncertainty visible until it has been resolved.

The case: updates not consistently implemented

The FDA’s September 2026 early alert on certain Abiomed Automated Impella Controllers describes hardware, firmware and software updates that had not been consistently implemented. A retrospective servicing-record review identified units requiring action, and Abiomed’s 18 September customer letter outlined coordination of removal, updating and return. The FDA notice also addresses continuity of care while service is arranged (U.S. Food and Drug Administration, 2026).

This article concerns the organization and verification of corrective work. It does not advise hospitals or clinicians to change device use or clinical practice. Those decisions must follow current official instructions and the responsible clinical and technical teams.

Follow the unit, not just the customer account

A hospital can acknowledge a notice without every affected controller being located. A service visit can occur without every unit on the original list being available. A device can move between departments while a central register still shows its previous location. These are hypothetical implementation gaps to test, not additional findings about this case.

The practical question is whether the campaign can trace an individual affected unit across organizational boundaries. Who owns the next action at the manufacturer, at the facility and within the relevant service function? What happens when the expected unit cannot be located?

One accountable coordinator can help, but responsibility should not depend on one person’s memory. The process needs a shared record that survives a shift change, staff absence or device transfer.

Define the evidence for every status

For a management review, I would distinguish at least four states:

• Identified: the affected unit and its present location have been matched to the campaign.
• Planned: an appropriate service arrangement is agreed, with responsibility for outstanding dependencies.
• Implemented: the specified corrective work is recorded against that unit.
• Verified: the evidence meets the completion criteria established by the responsible technical and quality teams.

The precise criteria must come from the relevant technical process; a marketing or management framework cannot supply them. The contribution of this distinction is to prevent an administrative milestone from being mistaken for a technical one.

I would also retain explicit exceptions: unit transferred, not located, service deferred under the applicable process, incomplete record, or correction awaiting verification. An exception is a task to resolve, not a convenient way to reduce the outstanding population.

Make the handovers part of the customer experience

My work with Shishi Liu proposes examining recall management across the customer journey (Raithel & Liu, 2026). Applied here as an analytical lens—not a tested clinical intervention—it encourages a review of what the institutional customer needs at each handover.

For example, does the facility know who will arrange the next step, what information to prepare and how a change in circumstances should be reported? Does the manufacturer receive an unambiguous confirmation, or an email that merely says the message was forwarded?

Clear communication can support execution, but it cannot replace the engineering and clinical safeguards. The service process must accommodate those safeguards rather than promise speed without regard to them.

Ask for an exception report, not just a completion percentage

Manufacturers and institutional customers can reconcile unit-level records and investigate mismatches. Law firms can examine whether claims of completion are supported by the underlying documentation. Insurers can consider how open actions and transfers are tracked, alongside—not instead of—technical risk assessment.

The early alert is not a complete account of the underlying causes or the eventual campaign outcome. It would be inappropriate to assign all responsibility to either the manufacturer or its customers from that notice alone.

A defensible completion figure starts with a defensible definition—and a visible account of the units that do not yet meet it.

References

Research scope: The public chapter abstract was consulted for the customer-journey perspective; no claims are made about evidence beyond that abstract.