An example of risk analysis of class I MD

[Watchcat]

So I finally got curious and took a look. I found this document useless from a regulatory perspective, because, unless I missed something, it doesn't identify the harms associated with the risks.

Engineering associates are typically not qualified to make a determination as what types of injury require professional medical intervention.

Oh good, my second opportunity to rant about this today. IMO, anyone really qualified to make this determination would tell you that trying to quantify the severity associated with a clinical harm makes no clinical sense whatsoever In this case, it's just plain weird, because you don't even know what harms are being rated. It's like, well, if some kind of harm occurred, then whatever kind of harm it might be, it would have a severity of 2.

(Feel free to view this as a potential audit finding against 13485 - Human Resources)

Don't tease me, Seymour! Have you ever actually known an ISO 13485 auditor to cite this as a finding?

 

[Ronen E]

Engineering associates are typically not qualified to make a determination as what types of injury require professional medical intervention.

As a boilerplate critique, I accept its validity (especially with the "typically" emergency exit included). But in terms of adding value (or of process effectiveness and process efficiency), and as a generalised approach, I think that it's overreaching. It all depends on the general risk realm the specific subject device is in, and the specific harms. Sometimes it's quite obvious and one doesn't need to be a MD or RN to tell. For borderline cases, consulting the right professionals is definitely warranted.

 

[Tidge]

For the example provided I have strong opinions where it could be improved, but I tried to restrict my comments to where I felt it was 'obviously' deficient. Watchcat's valid point about the way harms are (not) addressed is on the same spectrum as my complaint about who is capable of assessing the severity of the harms. My experience has been that a successful approach is to have an experienced medical professional take a comprehensive list of harms and then rank them according to a scale (typically 1 to 5) that allows their use in making risk assessments. If an organization has such a "Master Harms" list, it removes this element of subjectivity from individual risk management plans.

Not every harm from a general "Master Harms" list is going to be relevant for every device, but if the organization suddenly gets of complaint that includes a patient or user harm that wasn't originally considered in the risk file an organization with a general master harms list is more likely to be able to respond in a more mature (less reactive) manner than one that didn't even consider the possibility of that harm. For example, it might seem absurd to consider "choke hazards" for a corrective contact lens, but that doesn't mean a lens manufacturer could ignore a complaint that someone choked on one of their lenses.

People can quibble about the (possibly perceived as arbitrary) numerical value associated of such an assessment, but this is the only practical way I know of to address risks to apply a methodology that allows for non-serious harms to be less acceptable risks than those from more serious harms (S x O). An organization may have ranked "Finger cut that doesn't require stitches" as a S=2 and "Kidney necrosis" as an S=5, but if the users is cutting their finger every time they touch the device, this is almost certainly an unacceptable risk, independent of the risks associated with tissue necrosis.

 

[Ronen E]

My experience has been that a successful approach is to have an experienced medical professional take a comprehensive list of harms and then rank them according to a scale (typically 1 to 5) that allows their use in making risk assessments. If an organization has such a "Master Harms" list, it removes this element of subjectivity from individual risk management plans.

I don't disagree - my experience points in the same general direction. However, this falls (IMO) under the "if we lived in a perfect world... however..." advice category. Not every little design change process warrants engaging a (relevant) clinician, and one is many times not available within the org. When they aren't, it would usually cost not a few $$. the more complicated/risky the application is, the harder it will be to locate a relevant individual, and that individual's service is likely to cost more. Please don't forget that the clinician is taking a serious responsibility in doing so, especially if their output is "Masterised" and subsequently used in instances and contexts that the clinician can't predict. So aside from them jacking up the price (high risk warrants high return, that's Economics 101), they are also likely to jack up their severity assessments. An employee of the company (e.g. an engineering associate) is not personally exposed in that manner and thus likely to be less concerned.

BTW, the fact that an individual is a clinician and not an engineer doesn't mean that their judgement is not subjective. Less invested maybe (if they're external to the project), but not less subjective. The way I know to deal with subjective judgement is to obtain multiple opinions, but that makes the exercise even more costly (in some case it's justified).

this is the only practical way I know of to address risks to apply a methodology that allows for non-serious harms to be less acceptable risks than those from more serious harms (S x O).

I assume you meant "this is the only practical way I know of to address risks to apply a methodology that allows for non-serious harms to be more acceptable risks than those from more serious harms (S x O)."

People can quibble about the (possibly perceived as arbitrary) numerical value associated of such an assessment, but this is the only practical way I know of to address risks to apply a methodology that allows for non-serious harms to be less more acceptable risks than those from more serious harms (S x O).

I think that not many argue against risk ranking in this manner. I don't - as long as everyone is aware that the ranking is relative and subjective in most practical cases (where a lot of hard data is already available, that's great, but it's not so common in device D&D). I was arguing against getting overly wrapped up in "precise" calculations based in most cases on wishful thinking and a lot of coffee.

 

[Tidge]

I assume you meant "this is the only practical way I know of to address risks to apply a methodology that allows for non-serious harms to be more acceptable risks than those from more serious harms (S x O)."

I actually meant what I originally wrote, but perhaps I ought to have qualified the sentiment as "before the application of risk controls." It can be unacceptable to allow a near-constant (high occurrence) amount low-level harm (i.e. non-serious, low severity harms such as minor cuts, burns, etc.) just as it can be acceptable to have an extremely low occurrence rate of some very serious harms. This is of course, modulo discussions centered on "all risks shall be reduced as much as possible".

 

[Watchcat]

Sometimes it's quite obvious and one doesn't need to be a MD or RN to tell.

Can you provide an example?

 

[Ronen E]

I actually meant what I originally wrote, but perhaps I ought to have qualified the sentiment as "before the application of risk controls." It can be unacceptable to allow a near-constant (high occurrence) amount low-level harm (i.e. non-serious, low severity harms such as minor cuts, burns, etc.) just as it can be acceptable to have an extremely low occurrence rate of some very serious harms. This is of course, modulo discussions centered on "all risks shall be reduced as much as possible".

Thanks for clarifying, I think I understand now.
I agree that harm severity should not alone drive the acceptability determination. I didn't notice the subtlety you referred to because you specifically wrote "S x O". Both severity and likelihood should be taken into account when considering a risk.
Again, I'm all in for rating and ranking, as long as we recognise our own limitations and don't delude ourselves that its perfectly rational / objective / watertight or the likes. There is a lot of potential value in the process, but it's mostly in taking the time and attention to ponder and to look for a better understanding, not in the mechanistic application of a mathematical algorithm.

 

[Ronen E]

Can you provide an example?

The equivalent of a paper cut...?

 

[Watchcat]

Good example. How would you rate the severity of the clinical harms that are potentially associated with a paper cut, prior to application of risk controls?

 

[Ronen E]

Good example. How would you rate the severity of the clinical harms that are potentially associated with a paper cut, prior to application of risk controls?

If you recall, I was referring to making a determination as what types of injury require professional medical intervention. Not the severity of the clinical harms in general.