Estimation of risk reduction prior to verification of effectiveness

ThatSinc

Quite Involved in Discussions
Hi All,

I'm looking for some guidance on estimation of risk reduction (and therefore residual risk) prior to obtaining formal verification of effectiveness through design V&V, and the expectations/practical application of documenting the information that's been used to support the estimation.

For an invasive (body orifice) device where the dimensions & surface finish are significant factors in whether it stays in the intended location and performs as intended, whether it is able to move, and cause harm, or whether it is too large and will cause harm or is generally uncomfortable (low severity, potentially acceptable harm) due to surface finish - the dimensions and surface finish are the risk controls in place to mitigate this harm.

The verification of effectiveness of these controls is captured as a part of clinical investigation of the device.

The dimensions and surface finish are not subject to any national/international standards - they are defined by the biology of the intended user population, historical use, and knowledge by device designers with significant industry experience.
However, within any particular development project there are no records prepared to detail the above, or where the decision for any particular design outputs has come from - when senior developers are asked the response of "I've been doing this for X years, I know the sizes needed for the users".

Prior to performing the clinical trials, and obtaining specific evidence for the device in question, what recorded evidence would you expect to cite in the risk file for supporting the estimation of residual risk?

Thanks in advance,

TS.
 

DanMann

Quite Involved in Discussions
If I'm reading ISO 14971 correctly, you shouldn't need to estimate residual risk prior to verification of effectiveness; can you see a particular reason for this?
If you really need to, how do you justify the Design Inputs for the design of these features, as I expect this would be a similar answer.
Alternatively, if the developers have been doing it for years, could data from previous similar designs be used to estimate the residual risk prior to verification?
 

ThatSinc

Quite Involved in Discussions
To your first point - regulatory authorities have stated it's necessary in order to approve a clinical trial, despite the clinical trial being the ultimate verification of effectiveness. Though it's appreciated that if there's potential for harm, performing a clinical trial without evidence to support the decisions made up until that point isn't acceptable in my opinion.

To your second point - "I've been doing this for X years, I know the sizes needed for the users" - Verbal communication, Senior Clinician.
This is a concern for the development process documentation and records, manifesting itself in the risk management process.
 
Last edited:

Tidge

Trusted Information Resource
Without respect to the specific procedures, risks or terminology... I offer the following guidance when qualitative scales for occurrence and/or detectability are in use. To repeat: This is what I suggest my teams do to per policy or plan (because we don't yet have objective evidence of any sort:
I'm looking for some guidance on estimation of risk reduction (and therefore residual risk) prior to obtaining formal verification of effectiveness through design V&V, and the expectations/practical application of documenting the information that's been used to support the estimation.

Imagine that the qualitative scale has some ordinal ranking for occurrence (this is where I prefer to allow preliminary risk reduction, YMMV) such that the highest ordinal ranking (e.g. '5' on a scale that includes '1', '2', '3', '4', '5') is commensurate with "that risk is almost certainly going to manifest if we don't do something."

I will defend a preliminary risk reduction by multiple (ordinal) orders (usually a max of two, but I'm flexible) if the specific risk is addressed by a consensus standard that requires a (well-recognized, or obvious) implementation and testing against the standard. There are multiple examples, but 'knowing in advance' that the electrical device has to have a fuse for certain risks is something of a no-brainer. Keep in mind, one purpose of a preliminary risk assessment is to motivate design efforts. See also sterilization. These risk controls have to be verified as effective of course, but it isn't as if MORE risk controls need to be investigated for these sorts of controls.

If there exist always-used methods (at the company) in the process for design development (e.g. practices for PCBA layout) that exist and are used for specific risks (for PCBA layout, these risks could be EMI/EMC) I will defend a preliminary reduction of one order (of occurrence).

If the risk control is otherwise simply something like a 'design input requirement', I don't usually allow a preliminary risk reduction, as there is no value in arguing over preliminary risk reduction of individual DIR (in the risk files) since they all have to be verified to be effective anyway, and DIR can change with some frequency.
 

d_addams

Involved In Discussions
As Tidge said, there is no point in estimating 'risk reduction' on the initial design just estimate the risk given the design and the specifications (mitigations) originally developed. If the estimated risk can not be considered acceptable (as low as possible and benefit outweighs the risk) then additional mitigations are necessary. At that point you could then show post reduction estimates.

Regarding 'I've been doing this for X years'. Thats not an acceptable answer. In your stated question you said the design performance is a function of dimensions and surface finish, but the 'design' engineer's response didn't mention all of these aspects so it seems they think it is only a function of size. In the end there needs to be some assessment that the product is going to not fail to achieve its defined function (assuming in this case the function is 'remain in place'). So there needs to be requirements (specifications) that mitigate the risk of the loss of function (i.e. the product migrates after being put in position). This could be done with human clinical, animal tests, or bench tests (such as force to move after 'installing' the product in a simulated patient). This would then be a specified performance on a defined test. Specified performance on that test is the mitigation and demonstration of compliance to that requirement is your verification of effectivity of the mitigation. You'll need to be prepared to explain to a regulator why that test is sufficient to predict in-vivo performance. This could be based on competitive product testing (these 3 products are in the market and perform between X-Y on this test thus we have used those values to set our specification) or correlation between in-vivo testing and bench testing (takes X force to move the product in-vivo, takes Y force to move the same product in our test).
 
Top Bottom