Don't: Use Detectability to Estimate Risk to End-Users.
The above image, while obviously not a medical device, shows the potential foibles of using detectability to estimate risk to end-users.
Detection is a risk control measure—not an element of risk. Do not use detection in your device risk analysis, which is intended to estimate the risk to your end-users, once the product has gone out the door and is in the user’s hands.
Once the device is in the hands of end-users, manufacturers have no knowledge or control over what each individual end-user can or cannot detect. Even if an end-user detects a hazardous device defect, manufacturers can have no knowledge or control over whether the user will remember how to properly respond to the risk. And, even if the user recalls the correct response, manufacturers can have no knowledge or control over whether the user will have adequate time or inclination to do so.
The risk control, detection, is only useful to manufacturers before the device is shipped to the end-user. During manufacturing, properly trained inspectors and assemblers have an opportunity to catch manufacturing and packaging defects; they rarely have an opportunity, or the skill level, to catch design defects. If you want to mitigate a design risk, you need to do it long before it ever gets to the end-user.
—G.M. Samaras of Samaras & Associates, Inc. (Pueblo, CO)