Medical Device Verification and Validation: Stages, Costs and Design Controls

Medical device prototyping is the same engineering as any other prototype, plus a paper trail that has to hold up years later. When our team designed and built the UpShield face shield during the 2020 PPE shortage, the parts were printed in days - the work that mattered was material selection, cleanability and documenting why each decision was safe. Every medical program runs on that same balance of speed and evidence.

Prototype materials should be chosen with the final material in mind. Printed photopolymers are excellent for form and usability work but are almost never acceptable for patient-contacting verification units. Machined or molded medical-grade polycarbonate, PEEK, ABS, silicone and 316L stainless give you parts that can be cleaned, sterilized and tested against the same standards the production device will face. If a prototype will touch a patient in a study, plan for a documented material with a master file, not whatever resin was loaded that week.

Design controls (21 CFR 820.30, ISO 13485) do not require you to stop iterating - they require you to know which build you are talking about. In practice that means a revision-controlled drawing package per build, a risk file (ISO 14971) updated whenever the design changes, a design history file that captures why alternatives were rejected, and traceability from user needs through requirements to the test that verified each one. Teams that start this at the works-like stage ship faster than teams that reconstruct it before submission.

During the COVID-19 PPE shortage our team designed the UpShield, a secondary barrier worn over existing protective gear, and moved it from sketch to distributed parts in weeks. Three things made that speed possible: a geometry deliberately designed for one manufacturing process, a material chosen for cleanability before aesthetics, and a small team that documented each revision as it printed rather than afterward. Those habits transfer directly to a regulated device program.

Verification asks whether the device meets its design inputs; validation asks whether it meets user needs in the intended environment. Teams conflate them and then discover, late, that a device passing every bench test still fails a usability study. Plan both as separate protocols with separate acceptance criteria, separate sample sizes and separate reports. Verification can run on production-equivalent units built with production tooling; validation needs the real use environment, real users and the real labeling and instructions the customer will receive.

Sample size is a statistics question with a schedule consequence. Attribute tests (pass/fail) with a 95% confidence and 95% reliability target need 59 consecutive passing units; variable data with a known distribution can reach the same confidence with far fewer. Choosing variable measurement wherever possible is often the single biggest lever on verification cost, because it reduces both unit count and test hours.

Build the timeline backwards from the submission date and add explicit float for retest. Laboratories quote queue time separately from test time, and a failed EMC scan usually means a design change, a new build and a new queue slot. Teams that plan a single sequential path through testing are the ones that miss dates; teams that pre-book a retest window rarely do.

Verification asks whether you built the device to the specification. Validation asks whether the specification was the right one for the user and the use environment. Regulators expect both, documented separately, with traceability back to design inputs.

Sequence matters. Running usability validation before the design is frozen produces evidence that no longer applies to the shipped device, and repeating a human factors study is a six-figure mistake.

The most common V&V delay is discovering, mid-test, that the sample size or the acceptance criterion was never justified. Write the plan first and have it reviewed by someone who will not run the test.

Front-loading protocol review typically adds two weeks to the schedule and removes two to four months of repeat testing later - the best return available anywhere in a device program.

A concept model costs $2,000 to $8,000. A functional works-like prototype typically runs $25,000 to $90,000, and a production-intent verification build with testing reaches $60,000 to $250,000 depending on device class and electronics content.

Yes for concept, usability and non-patient-contacting fixtures, and increasingly for production in printed titanium and PEEK. Standard prototyping resins are generally not suitable for patient contact or sterilization without documented biocompatibility data.

Formally at design input, but practically as soon as a prototype informs a design decision. Keeping revision control and a risk file from the first functional build avoids reconstructing the design history file before submission.