Re: Validation Plan for a new Class II Medical Device (Critical Care Medical Ventilat
Apologies for the long digression but it is an interesting area. I found there are actually three distinct uses or meanings for word "validation", which often get mixed around:
1) field trials
2) the process of tackling complex situations
3) verification against top level specifications
Field trials are a great idea but they don't have much of a place in regulations per se as they are usually more of a fishing expedition (undefined, not final design). Manufacturers sometimes refer to field trials as "validation" but they really should be slotted in as part of the early prototyping, before specifications are fixed.
The "process" meaning is common in regulations, guides and also the meaning in ISO 13485. In this context, to say a system is "validated" means a complex system has been broken down into more manageable parts, verified, re-assembled systematically in a way which provides more confidence than just trying to tackle everything at a system level.
Finally there is meaning that validation is just top level verification, i.e. for the whole device as finally assembled, in the conditions of normal use. This seems to be the meaning in standards like IEC 62304 and IEC 60601-1. It is no different to normal verification, it is just the top level.
It is possible to combine meanings (2) and (3) and this is probably the best way, but it is not done in practice. An example helps to illustrate:
In the ventilator there is a flow sensor which is susceptible to room temperature. So there is separate sensor to measure the room temperature and software compensation is applied. You now have two options:
(a) at a lower level, verify the compensation works properly from 10°C to 40°C with detailed tests at various temperatures, flow rates and pressure. It could even involve further sub-system breakdown, hardware, software, ambient sensor parts and so on. Once complete, full system tests are done on a spot check basis only,
or
(b) full system tests need to be performed at 100's of combinations (room, flow, pressure settings) to make sure the system works under all conditions.
In principle either (a) or (b) is OK. But obviously (a) is more efficient and popular way.
The mistake though is to refer to the system spot checks as "validation". In truth, the validation is the combination of the spot checks, the lower level tests, and the decision that these combined are enough to provide confidence in the system.
There is a third option which is just buy an off the shelf sensor from a well-recognised maker with integrated compensation which means you don't have to worry about the lower level testing. Which illustrates the point that you really can't start writing detailed verification and validation test plans until the design is stable. Decisions on what to verify and validate will depend as much on the implementation as the "user specifications".
So, in the early stages, a "validation plan" should just be a placeholder in the overall design plan to say that validation plan will be developed when the system design is stable.
