Medical Device NPI Checklist

After the last gate

Launch is where NPI ends and two other processes take over. Every change to the device, its labeling, its suppliers or its manufacturing now goes through design change control, including a check on whether the change needs a new submission. And the post-market surveillance plan starts collecting the field data that will confirm, or challenge, the risk estimates made before launch.^[1][3]

Hold a short retrospective a few weeks after the first shipment, while the details are fresh. Which checks were done late, which were done twice, and which were missing from this list? Fold the answers into your own gate template before the next project starts.

The plan

The Tavora is a sterile, single-use scope for looking inside the bladder in a urology clinic, paired with a reusable handheld display that shows and records the image. Each scope is built by a contract manufacturer, packed in a sealed pouch, sterilized with ethylene oxide (EtO) and thrown away after one procedure. The display is charged, wiped down and used again.

That mix made the Tavora a typical new product introduction (NPI): electronics and software from the development team, a molded and bonded scope built by someone else, a sterile barrier and a shelf life to prove, labeling with unique device identifiers (UDIs), and a 510(k) in the US followed by CE marking in the EU. The plan gave each of those a lane, with a US launch in September 2025.

Here's how the plan and reality compare. The numbered markers match the sections below.

1 · January 2025 · Design freeze

The freeze that wasn't

The design freeze gate passed “with conditions.” Two items were still open: the adhesive for the scope's bending section, and the size of the sterile pouch. Neither looked serious. Both were written into the gate minutes, and neither had an owner or a due date.

The pouch was the one that hurt. In April, packaging engineering moved to a larger pouch so the scope's cable wouldn't kink. A sensible change, but the sterilization validation, the packaging validation and the shelf-life aging all depend on the final pouch, and none of them could start until it was settled. Nobody at the gate had asked what the open items could invalidate downstream.

2 · March–July 2025 · Design validation

Validated on the wrong scopes

To save time, the team ran summative usability testing and its design validation on scopes hand-built in the engineering lab. The contract manufacturer's first production scopes arrived a month later, and urologists who tried them noticed straight away: the bending section was stiffer.

ISO 13485 requires design validation on product that represents production, such as early production units or an equivalent, and the team must record why its choice qualifies.^[1] The team had no rationale that would stand up, because the units weren't equivalent. Part of the validation had to be repeated on production scopes, which cost five weeks.

3 · May 2025 · Pilot build

A supplier change nobody announced

During the first pilot build, some scopes came out of sterilization with hazy edges on the image. It took three weeks to trace: the camera module supplier had switched the adhesive that holds the lens, a change it considered minor. The new adhesive didn't tolerate the EtO cycle as well as the old one.

The purchase agreement didn't require the supplier to report changes. Where it applies, ISO 13485 expects a written commitment from the supplier to warn the manufacturer before making any change that could affect whether its part still meets requirements.^[1] Reverting the adhesive, requalifying the module and rebuilding the pilot lot took seven weeks in all.

4 · June 2025 · Design transfer

Transfer treated as a handoff

The plan showed design transfer as a single box near the end: hand the drawings, specifications and test methods to the contract manufacturer, and start building. In practice, the leak test that checks every scope's seal failed its first gauge study on the manufacturer's line. It had been developed in the engineering lab on a different fixture, by the engineer who designed it.

ISO 13485 expects a team to show, before release, that its design outputs can be built to specification on the production line, and that the line can do it consistently.^[1] A test method that only works in one person's hands isn't ready. Redesigning the fixture and repeating the study took three weeks, mostly hidden behind the supplier problem.

This kind of problem doesn't always stay hidden. When FDA reviewed 437 Class II recalls from October 2019 to September 2020, the most common root cause was process control: how devices are made, rather than how they were designed.^[55]

5 · July–October 2025 · Sterilization and shelf life

Sterile, but for how long?

Because the pouch changed late (section 1), the EtO validation started two months behind plan. The first runs showed EtO residuals above the team's acceptance limits, which are based on ISO 10993-7.^[28][32] The larger pouch and the denser shipping carton slowed aeration. Longer aeration fixed it, but the validation had to be partly repeated.

The shelf-life plan had a second problem. Accelerated aging, which uses heat to simulate time, can only start once the final product and sterile barrier exist. ASTM F1980 describes the method, and ISO 11607-1 accepts accelerated aging as evidence for the sterile barrier's expiry date while real-time aging continues. The device's own stability needs its own evidence.^[33][14] With the late start, an aging temperature limited by the lens adhesive, and only the data in hand by the time of the 510(k) response, the team could support just a six-month expiry at launch instead of the planned two years. Nobody had told the commercial team, who had planned inventory around two years.

6 · July 2025–January 2026 · 510(k)

Ninety days isn't three months

The plan assumed clearance three months after submission. FDA's review goal for a 510(k) is counted in FDA days, and the clock stops while a submission is on hold waiting for answers.^[56] The Tavora went on hold about two months into review, with an additional information request covering EtO residuals and software documentation. A submission on hold has 180 calendar days to respond; the team took eleven weeks, partly because the repeated sterilization work in section 5 meant new residual data had to be generated.^[56]

Clearance came in mid-January 2026, and the first commercial shipment left in February. Until clearance, the team could build inventory but couldn't sell or ship it commercially.

7 · February–March 2026 · US launch

Ready to ship, not ready to support

In the first week, three clinics called about the display freezing mid-procedure. The calls went to sales, who treated them as tech support: they talked the clinics through a restart and moved on. The complaint handling procedure had been written with the single-use scope in mind and didn't mention the display at all.

The gap surfaced a month later, in an internal audit. Every one of those calls should have been logged and evaluated as a complaint, including whether it needed a medical device report.^[1][44][49] They were logged late, investigated (a firmware timing fault) and judged not reportable, but the late records became an audit finding.

8 · April–July 2026 · EU launch

A regulatory assumption that went out of date

The EU launch slipped behind the US one while the notified body finished its review. By the time the certificate arrived, the rules had changed under the plan's feet. Since 28 May 2026, the first four EUDAMED modules have been mandatory, and a new device must be registered in EUDAMED before the first unit is placed on the EU market. There's no transition period for new devices.^[48][3]

The launch plan, written in 2024, had assumed registration could follow later. Registering the device in EUDAMED, with its Basic UDI-DI and UDI-DI data, added three weeks before the first EU shipment in July.

What added up to five months

None of the eight problems was a delay of five months on its own. They stacked: an open condition at design freeze moved the pouch, the pouch moved sterilization, sterilization moved the submission and the expiry date, and the hold moved clearance. Meanwhile, the supplier and test method problems ate the slack that might have absorbed the rest.

In their retrospective, the team agreed on a short list of changes for the next product:

Questions worth asking at your next gate review

Copy these into your gate template and see how many your team can answer with evidence, not opinion. For the full list, use the gate-by-gate checklist above.