Design output includes detailed specification, therefore some output need to be validated by testing. In some cases, medical device cannot be made by R&D, and samples provided by manufacturing will give more confidence about validation results. That's what I'm understanding. For severity and confidence level, it depending on company's experience and customer complaints

James (Jim) Dent, LSSBB, DTMx2 Are you confident your product is safe and defect free based on reliability and confidence data; or just a persona opinion based on measuring one sample item?

"Without data, you're just another person with an opinion." -- Deming

The need for reliability and confidence factors based on real sample data, and how to collect and determine your reliability and confidence levels is described in Juran's Quality Control Handbook. Which level of R/C you seek is left up to you based on the risks. Will a failure of the part result in death, or just be a fairly harmless nuisance.

R & D are great at making products once in a row in a lab. You need to run production scale qualifications to assess operator influence, hidden defects that don't appear in prototyping because of a low occurrence, R & R of multiple production lines and sites.

Karen Boyd, ASQ CQA My thoughts align with Feargal, in that design for (or from) Engineering may not always meet design for manufacturability.

Dan O'Leary My question is simple. Design verification confirms that the design output meets the design input requirements. How did this morph into manufacturability? To put it another way, if I have trouble manufacturing a product described by design output, I would expect to fix the manufacturing process.

The design input/output specification is/was incomplete without manufacturability. "On paper everything works".

James (Jim) Dent, LSSBB, DTMx2 Design verification / validation requires that your design output meets your design input. Your design input specification may be much more than just material and feature dimensions; your design input may include:
* useful life

* 99% reliability (a risk mitigation factor)

* 95% confidence (a risk mitigation factor)

* XX amount of wear on surfaces X and Y over the life of the product

* re-sterilization using only steam sterilizing at the clinic/hospital/ medical center

* a shelf life (in the original package) of XX years, under normal storage environments

Therefore sample size is dependent on what your design input requirements are for the particular product. These requirements can differ from one product to another.

I agree with Dent, Zamfir, Judge in particular, and all general. The design model on the input may not represent the manufactured output due to hidden defects, or unanticipated variables that are not modeled in the original design. You must run pilot builds of subassemblies, and complete assemblies in order to insure that the design input meets the manufacturing output validation measures, and that unanticipated variables are captured and modeled with appropriate confidence interval.

James (Jim) Dent, LSSBB, DTMx2 The design verification may not apply only to the product itself, but to the implanted application, if an implant - it's long term compatibility with the bone, tissue, it's bonding with bone and tissue. Is it designed as a temporary trauma device, provisional temporary implant (less than 30 days), or an implant intended to remain in the body for 10+ years, or is it a laboratory instrument for testing tissue samples in pathology? The end-use is also part of the design input.

Dan O'Leary Thank all you for the interesting comments. I appreciate your taking the time.

The comments have, again, helped me clarify the question. In either FDA QSR or ISO 13485:2003 I don’t see any requirement for manufacturability as part of design verification. While manufacturability is important, I don’t believe it properly belongs in design verification.

If commenters believe it does, I think it would be useful to say why. This would help distinguish between good practice and a regulatory requirement. In particular, if design verification did not include manufacturability, would that be grounds for an FDA Warning Letter?

In design, in addition to the materials being selected the tolerances applied to a part / assembly must take into account the likely manufacturing route(s) for a particular process or treatments.
The end of the design cycle is Design Validation from 'initial production units or similar' FDA 820.30. If the verified design did not take account of manufacturing variabilities it would make this a step a high risk and unlikely to succeed as the initial production units would possibly not meet their release requirements.
Not considering manufacturing as part of design is poor design practice.

Dan, I think your question is in place. Indeed design verification purpose is to show that design output meets the design input and manufacturing process is not part of it. However, FDA recommend to make FMEA in order to identify possible risks as a results of poor design or bad manufacturing quality or process. As a result, risk assessment should perform and mitigation should take either inherit design, labeling or quality process during the production. In some cases, the only way to verify the risk mitigation is by verification post process validation with a final product.
If you have a comprehensive management file that shows that manufacturing process not make new risks and/or also not involved risk mitigation that was not verified, you can be confident that you will not get any warning letter. The risk management FDA guidance can clarify few points for you.

Farhan Mahmood I agree with Itzik. When dealing with Class II and III medical device, design for manufacturability is crucial.

Antonin Cuc Dear friends, I have many experiences with industrial statistic testing all products, all components of products, all operational partial testing each elementary task on each workplaces in the technologic prescripcions of processing with described measurement instruments, tolerancy of measurement, Paln of elementary mandatory instruments in the sequential processing of control with firm strategic style of usage statistic method with changes of accuracy and reliabilities....by the firm Standard of statistic acceptance of main firm production....for example in the firm TESLA, tools machinary productions in TOS....But the tragedy there are "to shorten Transformal Shannon´s Channel as definition to product perfect Medical Devices only", when then is continual Phases of Usage Medical Devices in daily medical workflow without real supervised monitoring extremal frequency of medical users mistakes contrary safety Technician requirements of Law...and Mass of Physicians don´t understand the bariers in informal flow between interfaces TECHNICIANS EVIDENCES LEGAL WORKING and Medical repeated Mass frequences of trivial user´s mistake with extremal patient risk. The Physicians didn´t understand there are full illegal to use "intuitive method medical working" in situation, where there are mandatory technician requirements from product instuction...The Physicians crimi activities are produced as Mistakes of God´s, but there are many thousands dying or injured patients....there are no technician controlling supervision...there ar no criminal guarancy of safety usage in medical workflow...I am as a State investigator Health and Safety daying...when the orthopaedic surgery THA began without mandatory preliminary Planning of used components in centric anchoring position in Hip bones...the Orthopaed began cutting "by the 20 years surgeon praxis" full false, when there are should detect my dysplastic declination of femoral neck...it causes Crash "Fausse route stem", but the Orthopaed realised functional testing "by the way and intuitive only", the first postoperstional image of RTG was only 1...but "with intuitive observing interpretation only" it was full out of product instruction...when there must be on orthopaedic screen comparing the mandatory Profile and Dimesnion by the firm radiological sagittal etalon...the false declination of stem from coaxiality of femoral bone was in the same direction of view as RTG emiter....Just I am dying on growing Health complication...the Physicians are sure "it was standarding medical processing" and the Criminal Policy refused to detect the Mass illegal medical workflow in the Orthopady and extremal false medical Court message "Lege Artis" and I am coauthor of harmonised Law EU/Czech in Directive 93/42/EEC Medical Devices! Just I should asked the help from some Criminal Anthropology Labour from OECD...and we all are sure "Such illegal ways are yearly thousands of patients needless dying out of Constitutional Rights to have the Safety Health care. to live and to keep the justice technical evidences in Court processing with detection biostatistic probably extremal shorten prediction of my Life! This is causes why I am recommending to derive abou 2% of all yearly Mass from EHRs databases to regularly criminal controlling guarancy of technician legal medical working in each Hospital OECD...there must be sharing data about most dangerous repeated strategic medical mistakes in typical medical Tasks, for example "THA Classic surgery" with continual Videorecording processing on surgery Hall, the right usage of assembling method and testing of orthopaedic products in duration of surgery, etc. etc...with sharing the best latest experiences with minimize patient risk,...allways with minimize summary increases entropy in defined Complex long being Transformal Channel, Utility model 21532 Czech Republic.

Interesting. I agree completely with Dan, but there is a very big confusion about this in the industry, and as also seen in some of the comments. My stand point is that if you are testing during design verification, you are doing something wrong (or at least not optimal).