As Dan has indicated there is still a problem with your harm table. I believe you are using FMEA as a risk analysis tool, which it is not. FMEA is a reliability tool that is used to identify failure causes and effects. The severity identified in the FMEA is not related to patient harm but is a measure of the severity of the effect of the failure on the operation of the area of the product being analyzed. Your table you have proposed does not apply to the analysis you are performing. Dan has discussed severity in relation to the harm resulting from an exposure to a hazard in a hazardous situation. You are at the point where you are identifying failure effects that might result in a hazard. (See Annex E in ISO 14971). The incorrect use of FMEA in ISO 14971 has resulted in much confusion on the definition of risk and severity which are used in FMEA that are different from the definition of risk in ISO 14971. In fact, IEC 60812 did not define risk or severity where ISO 14971 is very specfic in these terms.
My recommendation to you is to scrap the proposed table and start over using the severity of the effect of the failure on the function of the section of the product you are analyzing. Some FMEAs are focusing on a subassembly while others are on the entire product or system. So you have to take into account the impact of the failure of a component of the section covered in the analysis.
Original Message:
Sent: 06-Aug-2018 21:52
From: Dan O'Leary
Subject: Risk Hazard Likelihood criteria
There is still a problem distinguishing between hazards and harms. It is harms that have severity and probability and not hazards,
The severity of a harm is not necessarily constant. For example, if the harm is a burn it might be first degree, second degree, or third degree. Each burn severity would have a different frequency of occurrence.
A risk reduction measure involving a burn might reduce the surface temperature of a hot surface. The hazardous situation could be "high temperate on an exposed surface". If the patient or user comes into contact, then harm results. Assume the harm is a burn, but now, because of the risk reduction, the severity is reduced to a first degree burn only. This risk reduction measure would probably not change the frequency of occurrence of contact with the hot surface.
You say. "When you determine a change to the likelihood criteria you need to determine the number of complaints arising to the number of uses." Remember that the likelihood is a rate (a ratio) with a numerator and a denominator. Initially, the risk management team made their best guess with all (perhaps limited) available knowledge. After product release, collect production and post-production information and use it to update the initial guess. If the initial guess is low, then update the risk management file and determine if the residual risk is still acceptable. For example, the initial guess might be 3 occurrences per 1,000 uses, but the post-production data shows it is 15 times per 1,000 uses.
It is usually practical to determine the number of uses; often marketing has good estimates. It may make some assumptions. For example, you might assume that a home use blood pressure machine is used one time per day; the patient does a measurement every day. However, some households have multiple users who share the same machine. Today, machines often record historical results based on the identification of 3 to 5 users. With marketing's help you can get a good estimate of the denominator. Complaints provide the numerator, but you need to consider the underreporting problem.
------------------------------
Dan O'Leary
Swanzey NH
United States
Original Message:
Sent: 05-Aug-2018 23:55
From: Anonymous Member
Subject: Risk Hazard Likelihood criteria
This message was posted by a user wishing to remain anonymous
Thanks Dan and Edwin for your responses.
Level
|
Severity
|
Definition
|
1
|
Negligible
|
Inconvenience or temporary discomfort
|
2
|
Minor
|
Results in temporary injury or impairment not requiring professional medical intervention
|
3
|
Significant
|
Results in injury or impairment requiring professional medical intervention
|
4
|
Serious
|
Results in significant permanent impairment
|
5
|
Critical
|
Results in life-threatening injury or death
|
|
Likelihood of Hazard Occurrence ranking
|
Hazard Severity ranking
|
1 Improbable
|
2 Remote
|
3 Occasional
|
4 Probable
|
5 Frequent
|
5 Critical
|
5
|
10
|
15
|
20
|
25
|
4 Serious
|
4
|
8
|
12
|
16
|
20
|
3 Significant
|
3
|
6
|
9
|
12
|
15
|
2 Minor
|
2
|
4
|
6
|
8
|
10
|
1 Negligible
|
1
|
2
|
3
|
4
|
5
|
Table –Risk Index Matrix
Table – Residual Risk Acceptability
Risk Index
|
Definition
|
1 to 9
|
Acceptable
|
10 to 25
|
Not acceptable
|
So we have a separate Hazard severity table and risk Index matrix . I assume the severity of a particular harm will remain constant and the likelihood can change. When you determine a change to the likelihood criteria you need to determine the number of complaints arising to the number of uses . Is it practically possible to determine the number of uses for multiple use equipment?
Original Message:
Sent: 03-Aug-2018 07:45
From: Dan O'Leary
Subject: Risk Hazard Likelihood criteria
You didn't state the context for this proposed table; I infer you are implementing ISO 14971:2007 or EN ISO 14971:2012. I have some concerns about this table.
First, the table is about Risk Hazard Likelihood, but it should be about harms. ISO 14971:2007 is a process standard which has specific and well defined steps.
Identify hazards, recognizing that a hazard is a potential source of harm and could exist in either normal or fault conditions. Normal conditions rule out the use of FMEA.
Identify a sequence of events related to the hazard. This is a very important step, because risk reduction measures should address the steps in the sequence.
Identify a hazardous situation which is the "state of the world" when the sequence of events occurs.
When a patient or user is exposed to the hazardous situation, harm occurs. Identify the harm.
Harm has severity and frequency of occurrence.
Combine the harm's severity and frequency of occurrence to estimate the risk.
These are the steps in Clause 4 of the standard.
The frequency of occurrence table addresses how often the harm occurs with the stated severity.
Your table looks more like a reliability FMEA, in which you are evaluating the frequency of occurrence of equipment failure instead of the patient or user harm.
My recommendation is that you start with Table D.2 in the standard which provides three qualitative levels. Expand it to five levels recognizing that you are describing the frequency of occurrence of the harm, not medical device failure. Use the terms from Table D.4.
The frequency of occurrence is a rate, which means it has a numerator and a denominator. It is very important that you define them; the definition depends on the device. For example, you might say "Harm X occurs with Serious severity 3 times in every 1,000 uses." The numerator is how often the harm occurs and the denominator is device use. (Single use and continuous use devices would have different denominators.)
Because this is a frequency of occurrence, a rate, it does not depend on the number of devices in use. The rate is a constant and doesn't change when the number of devices goes from, say, 100 to 1500. (The number of complaints will change based on the number of devices in use.)
------------------------------
Dan O'Leary
Swanzey NH
United States
Original Message:
Sent: 02-Aug-2018 00:55
From: Anonymous Member
Subject: Risk Hazard Likelihood criteria
This message was posted by a user wishing to remain anonymous
Hello,
Is anybody able to confirm whether the below table looks ok for Risk Hazard Likelihood criteria
Table - Hazard Likelihood
Level | Probability | Definition |
1 | Improbable | Highly Unlikely to happen during the expected service life per unit and has never occurred |
2 | Remote | Unlikely to happen during the expected service life per unit, but possible. |
3 | Occasional | Likely to happen once per unit during the expected service life |
4 | Probable | Likely to happen multiple times per unit during the expected service life |
5 | Frequent | Likely to happen during each use per unit |