Products a company must support for decades

The MDR required every manufacturer to revisit products that had sold for years under the former directives. Total hip and knee replacements are class III devices, so the clinical evaluation, post-market clinical follow-up and notified-body review are substantial. I lead the portfolio review that decides what we put forward. For each system we compare the cost of new evidence and testing, its sales trend, the surgeons and patients who depend on it and whether a current product can serve them instead. Regulatory colleagues estimate the certification work; I own the recommendation. Withdrawal is not abandonment. We plan the supply of components and instruments that patients already implanted may need, the training surgeons require to change systems and the years over which we will keep answering for a product we no longer sell.

Our recovery app is a class IIa medical device, and every screen a patient sees belongs to its risk file. I work with our clinical safety lead and human-factors team to decide what the app may tell a patient and what it must leave to a clinician. The rule I hold to is short: the app may tell a patient what to do next, but it may never tell a patient that a symptom is normal. Questions about calf pain, breathlessness, fever or wound discharge lead to the hospital's own telephone number or to emergency advice, never to reassurance. New wording is tested with patients in their seventies before release. When a hospital asks for a feature that would need its staff to reply, we ask who will reply, and when, before anyone writes code.

Usage data has been more honest with me than feature requests. In 2023 surgeons opened our clinician dashboard a median of twice a quarter, while physiotherapists used the exercise-adherence view almost daily. We stopped developing the surgeon dashboard and now place a six-week summary in the hospital record, where a surgeon already looks before a follow-up appointment. In theatre, I ask scrub nurses to count what is opened from each instrument tray. Our older knee sets carried pieces used in fewer than one operation in twenty that still had to be washed and sterilised every time; standard knee procedures now need six trays instead of nine. I treat poor adoption as evidence about the product, not as a training gap to close after launch.

Hospitals buy our implants through tenders and framework contracts, and pay for them within financing rules that differ in every country we serve. A recovery app has no automatic payer. I examined Germany's route for prescribable digital health applications, which requires evidence of a positive care effect, and concluded that our clinic-led programme did not fit it: the study would have cost about €1.9 million and the listing would have described a different product. We charge hospitals €38 per patient episode instead. That price only makes sense where a hospital values fewer avoidable outpatient contacts or a more predictable discharge, so our product managers learn how each health system pays for a joint replacement before they promise anyone a benefit. Where we cannot describe that benefit honestly, we do not sell the app.

The median age of patients in our programme is 69, and roughly one in six has no smartphone. I want the programme to work for them without making anyone feel they have failed a technology test. We offer a printed exercise booklet that mirrors the app, a weekly telephone check-in hospitals can run from the same schedule and a relative's account that the patient approves and can withdraw. Health data is special-category data under the GDPR, so consent, access and retention for relatives' accounts were designed with our data-protection officer from the start. Text size, contrast and spoken instructions are tested with patients a few days after surgery, when they are tired and sore, rather than with healthy volunteers in a meeting room.