Q9_Failure_Mode_Effects_A.ppt

Annex I: Methods & Tools
prepared by some members of the ICH Q9 EWG for example only; not an official policy/guidance July 2006, slide 1
ICH Q9 QUALITY RISK MANAGEMENT
Annex I.2
Failure Mode
Effects Analysis
(FMEA)

I.2: Failure Mode Effects Analysis (FMEA)
(see IEC 60812)
 Evaluation of potential failure modes for processes
 The likely effect on outcomes and/or product performance
 Once failure modes are established,
risk reduction can be used to
eliminate, reduce or control the potential failures
 FMEA relies on process understanding
 Summarize the important modes of failure, factors causing
these failures and the likely effects of these failures
How to perform?
Break down large complex processes into manageable steps
ICH Q9

Potential Areas of Use(s)
 Prioritize risks
 Monitor the effectiveness of risk control activities
 Equipment and facilities
 Analyze a manufacturing process
to identify high-risk steps or critical parameters
RNP: Risk Priority Number C. Kingery, The Six Sigma Memory Jogger II
ICH Q9

How to perform?
1. Establish a team
2. Identify the known and potential failure modes:
Develop lists of known problems and brainstorm other
potentials…
e.g.
> Product not meeting specification
> Process not meeting yield requirements
> Malfunctioning equipment
> Software problems
Newly identified failure modes should be added at any time
EXAMPLE

How to perform?
3. Characterise the severity, probability and detectability
 An equal number of levels is sometimes helpful
> Some preference to 3, 4, 5, 6 or 10 levels
> But: an even number of levels avoids the mid point
 Use different scales
> Linear: 1, 2, 3, 4
> Exponential: 1, 2, 4, 8
> Logarithmic: 1, 10, 100, 1000
> Self made: 1, 3, 7, 10
Multiplying different scales will differentiate the outcome
The aim is to come
up with a method
of prioritising
EXAMPLE

How to perform?
4. Define actions
5. Revisit the ranking
6. Define residual risk
7. Perform a short summary
> Scope
> Data from the assessment & control
(e.g. No. of identified failure modes)
> Level of accepted risk without actions i.e. residual risk
(e.g. Risk priority Number < 50)
> Recommended actions, responsibilities and due dates
(including approval, if appropriate)
> Person in charge for follow-up of FMEA
EXAMPLE

Severity (Consequences of failure)
• 10 Extreme
• Predicted to cause severe impact to quality (Product out of
specifications, no Expert Statement possible)
• 7 High
• Predicted to cause significant impact on quality (Failure to meet
specifications, no Stability data, Expert Statement possible)
• 3 Moderate
• Predicted to cause minor impact on quality (Failure to meet
specifications, Stability data available)
• 1 Low
• Predicted to have no/minor impact on quality of the product
(Quality within specifications)
EXAMPLE

Probability (Likelihood failure will happen)
• 8 Regular failures
• Expected to happen regularly
• 4 Repeated failures
• Expected to happen in a low frequency
• 2 Occasional failures
• Expected to happen infrequently
• 1 Unlikely failures
• Unlikely to happen
EXAMPLE

Detectability (Ability to find the failure)
• 4 Normally not detected
• Failure very likely to be overlooked, hence not detected
(no technical solution, no manual control)
• 3 Likely not detected
• Failure may be overseen
(manual control, spot checks)
• 2 Regularly detected
• Failure will normally be detected
(manual control, routine work with statistical control)
• 1 Always detected
• Failure can and will be detected in all cases
(monitoring, technical solution available)
EXAMPLE

FMEA: Quantitation of Risk : Severity
10 Dangerously High Failure could lead to death or permanent injury to the customer. Financial:
>$1,000,000
9 Extremely high Failure could lead to injury to the customer. Failure would create non-compliance
with registered specifications. Failure likely to lead to recall. Financial: $1,000,000
8 Very High Failure could lead to adverse reaction for customer. Failure would create
noncompliance with GMP regulations or product registrations. Failure possible to
lead to recall. Financial: $500,000
7 High Failure leads to customer percept ion of safety issue. Failure renders individual
unit(s) unusable. Failure causes a high degree of customer dissatisfaction. Recall
for business reasons possible but Authority required recall unlikely. Financial:
$100,000
6 Moderate Failure causes a high degree of customer dissatisfaction and numerous
complaints. Failure unlikely to lead to recall. Financial: $50,000
5 Low Failure likely to cause isolated customer complaints. Financial: $10,000
4 Very Low Failure relates to non-dosage form issues (like minor packaging problems) and
can be easily overcome by the customer. Financial: $5,000
3 Minor Failure could be noticed by the customer but is unlikely to be perceived as
significant enough to warrant a complaint.
2 Very Minor Failure not readily apparent to the customer. Financial: <$1,000
1 None Failure would not be noticeable to the customer. Financial: none
Dr. Gary Harbour, Pfizer
EXAMPLE

FMEA: Quantitation of Risk : Probability
10 Very High: Failure is
almost inevitable
More than one occurrence per day or a probability of more than three occurrences in
10 units (Cpk < 0.33 or <1σ).
9 One occurrence every three to four days or a probability of three occurrences in 10
units (Cpk ~ 0.33 or ~1 σ).
8 High: Repeated
failures
One occurrence per week or a probability of 5 occurrences in 100 units (Cpk ~ 0.67
or ~2 σ).
7 One occurrence every month or one occurrence in 100 units (Cpk ~ 0.83 ~2.5 σ).
6 Moderate:
Occasional Failures
One occurrence every three months or three occurrences in 1,000 units (Cpk ~ 1.00 or
~ 3 σ).
5 One occurrence every six months to one year or one occurrence in 10,000 units (Cpk
~ 1.17 or ~ 3.5 σ).
4 One occurrence per year or six occurrences in 100,000 units (Cpk ~ 1.33 or ~ 4 σ).
3 Low: Relatively few
Failures
One occurrence every one to three years or six occurrences in 10,000,000 units (Cpk
~ 1.67 or ~5 σ).
2 One occurrence every three to five years or 2 occurrences in 1,000,000,000 units
(Cpk ~ 2.00 OR ~6 σ).
1 Remote: Failure is
unlikely
One occurrence in greater than five years or less than two occurrences in
1,000,000,000 units (Cpk > 2.00 OR >6 σ).
For batch failures use the time scale for unit failures use the unit scale. Dr. Gary Harbour, Pfizer
EXAMPLE

FMEA: Quantitation of Risk: Detection
10 Absolute
Uncertainty
The product is not inspected or the defect caused by the failure is not detectable.
9 Very Remote Product is sampled, inspected, and released based on Acceptable Quality Level
(AQL) sampling plans.
8 Remote Product is accepted based on no defects in a sample.
7 Very Low Product is 100% manually inspected in the process.
6 Low Product is 100% manually inspected using go/no-go or other mistake-proofing
gauges.
5 Moderate Some Statistical Process Control (SPC) is used in the process and product is final
inspected off-line.
4 Moderately High SPC is used and there is immediate reaction to out-of-control conditions.
3 High An effective SPC program is in place with process capabilities (Cpk) greater than
1.33.
2 Very High All product is 100% automatically inspected.
1 Almost Certain The defect is obvious and there is 100% automatic inspection with regular
calibration and preventive maintenance of the inspection equipment.
Dr. Gary Harbour, Pfizer
EXAMPLE

Severity / Probability / Detection (SPD)
PhD R.C. Mendson, Menson & Associations, Inc
ICH EWG London, March 2004
EXAMPLE

 Severity (S)
> Link to end product functional failure
> Medical Department involvement
 Probability (P)
> Use historical data
> Similar processes products
 Detection
> Method validation studies
> Historical data
EXAMPLE
Takayoshi Matsumura, Eisai Co.
Drying Process

Ranking Severity (S) Probability (P) Detection (D)
10 Death More than once a day Impossible to detect
9 ↓ 3 – 4 times a day Remote
8 Permanent injury Once a week Very slight
7 ↓ Once a month Slight
6 Temporary injury Once in three month Low
5 ↓ Once in half – one year Medium
4 Reported/ dissatisfied Once a year Moderately high
3 ↓ Once in 1 – 3 years High
2 Notice/ no report Once in 3 – 5 years Very High
1 ↓ Less than once in 5 years Virtually certain
Takayoshi Matsumura, Esai Co
EXAMPLE
Drying Process

Process Potential Failure Mode Potential Cause S P D RPN
1.
Set up
contamination disheveled gown of operator
insufficient cleaning of equipment
2.
Start drying
contamination damage of inlet-air filter
degradation of product damage of thermometer
3.
Maintain
temperature
long drying time unstable supply-air volume
high Loss On Drying
(LOD)
damage of timer
low LOD high dew-point
non-uniformity of LOD uneven temperature distribution
Drying Process
Takayoshi Matsumura, Esai Co
EXAMPLE
RPN: Risk Priority Number = S*P*D

Existing controls: IPC of LOD and degradation product after drying process
Drying Process
Process Potential Failure Mode Potential Cause S P D RPN
1.
Set up
contamination disheveled gown of operator 3 5 8 120
insufficient cleaning of
equipment
7 2 8 112
2.
Start drying
contamination damage of inlet-air filter 7 3 6 126
degradation of product damage of thermometer 7 3 3 63
3.
Maintain
temperature
long drying time unstable supply-air volume 2 4 5 40
high LOD malfunction of timer 2 2 2 8
low LOD high due-point 3 3 3 27
non-uniformity of LOD uneven temperature distribution 3 5 3 45
Takayoshi Matsumura, Eisai Co
EXAMPLE

Take action when RPN is over 100
Take action when severity is over 5
Remaining critical parameters after taking action; further controls required
Drying Process
Process Potential Cause RPN Recommended Action S P D RPN
1.
Set up
disheveled gown of operator 120 use long gloves and goggles 3 2 8 48
insufficient cleaning of
equipment
112 change cleaning procedure 7 2 4 56
2.
Start drying
damage of inlet-air filter 126 change maintenance period 7 2 6 84
damage of thermometer 63 change calibration period 7 2 3 42
3.
Maintain
temperature
unstable supply-air volume 40 ― 2 4 5 40
malfunction of timer 8 ― 2 2 2 8
high dew-point 27 ― 3 3 3 27
uneven temperature
distribution
45 ― 3 5 3 45
Based on Takayoshi Matsumura, Esai Co.
EXAMPLE

 Analyse a granulation process step
because only a few parameters are adjustable and many
problems can occur by manual operations
EXAMPLE
S. Rönninger, Roche
Severity (Consequences):
3: high Predicted to cause significant impact to quality (failure to meet specifications)
2: moderate Predicted to cause minor impact to quality (failure to meet specifications)
1: minor Predicted to could have minor impact on quality of the product (quality within specifications)
Probability
4: regular failures Expected to happen frequently
3: repeated failures Could happen occasionally
2: occasional failures Expected to happen infrequently
1: failure is unlikely Unlikely to happen
Detectability
3: probably not detected May overlook a fault or failture possibly can not be detected (no technical solution up to now)
2: occasionaly not detected Failture may be missed (manual control, routinely work with statistical control)
1: detectable Failture can and will be detected (e.g. using statistical tools)

Risk Assessment Risk Reduction
Sub-Step
Event
(Failure mode)
Effect
Severity
(S)
[1<2<3]
Probability
(P)
[1<2<3<4]
Detectability
(D)
[1<2<3]
Risk
factor
(S*P*D)
Actions:
Risk reduction strategy
Severity
(S)
[1<2<3]
Probability
(P)
[1<2<3<4]
Detectability
(D)
[1<2<3]
Risk
factor
(S*P*D)
Risk
reduction
Comments
Wet seving Drying Temperature
not meet specification of
degradation
2 4 1 8 implement 2 temperature measures 1 1 1 1 7
automatically interruption
by not meeting range;
Temperatur monitoring in
batch record
Granulation Drying water content
degradation
2 3 1 6 introduce online NIR 2 1 1 2 4 indirect measurment
introduce IPC analytic 2 2 1 4 2
direct measurement; time
consuming
humidity measurement in the exausting
air
2 1 2 4 2
indirect measurment;
unspecifoc
Granulation kneeding time
dissolution
3 3 1 9 reduce personnal fluctuation 3 3 1 9 0
operator knowledge;
depending on power
consumption;
automatisation not possible
at that time
Granulation power consumption
dissolution
3 2 1 6
try to get to a minumum an optimum of
kneeding time
3 2 1 6 0
depending on kneeding
time depending on material
properties
Pre-mixing mixing time
content uniformity
3 2 3 18 IPC measure on content uniformity 3 2 1 6 12 influence on efficacy
Pre-mixing Granulation speed of adding water
disolution and
desintegration
3 3 3 27
Analyse (seeving of granulate sieve
analysis); use of dosage pumps
3 2 1 6 21
to get fine appropriate
granulate
Pre-mixing Granulation manner of adding water
disolution and
desintegration
3 1 1 3 install spray nozzles 1 1 1 1 2
to get fine appropriate
granulate
Granulation Quality of Excipients
all parameters have to be
re-evaluated
3 4 3 36
Adapt internal specification of physical
parameters (e.g. density, metability
wetability)
1 2 2 4 32 contact supplier
Granulation Quality of API
all parameters have to be
re-evaluated
3 4 3 36
Adapt internal specification of physical
parameters (e.g. density, metability
wetatility)
1 2 2 4 32 contact supplier
Overview Risk before cotrol Max 36 Risk after control Max 9 32
Average 17 Average 4 10
Min 3 Min 1 0
Failure
Mode
Effects
Analysis
(FMEA)
EXAMPLE

I.2 Failure Mode Effects Analysis (FMEA)
Risk Assessment
Sub-Step
Event
(Failure mode)
Effect
Severity
(S)
[1<2<3]
Probability
(P)
[1<2<3<4]
Detectability
(D)
[1<2<3]
Risk
factor
(S*P*D)
Granulation Drying water content
degradation
2 3 1 6
Risk Reduction
Actions:
Risk reduction strategy
Severity
(S)
[1<2<3]
Probability
(P)
[1<2<3<4]
Detectability
(D)
[1<2<3]
Risk
factor
(S*P*D)
Risk
reduction
Comments
introduce online NIR 2 1 1 2 4 indirect measurment
introduce IPC analytic 2 2 1 4 2
direct measurement; time
consuming
humidity measurement in the exaust air 2 1 2 4 2
indirect measurment;
unspecific
EXAMPLE

 Prepare a risk profile
Severity / Consequences
i negligible
ii marginal
iii critical
iv catastrophic
Probability
A frequent
B moderate
C occasional
D rare
E unlikely
F very unlikely
Consequences
Risk
protection
level
EXAMPLE
Picture: © Zurich Insurance Ltd, Switzerland

 Prepare a risk profile: Probability
EXAMPLE

 Risk Evaluation
> Prepare a risk profile: Consequences
EXAMPLE
S. Rönninger,
Roche

 Risk Evaluation: Risk Profile
> For high risks, which are not acceptable, risk reduction
measures have to be taken as a high priority
EXAMPLE

How to perform?
3. Summary (Risk Evaluation)
> The effects are rated in terms of their
consequences and the
causes are assessed in terms of their probabilities
a) qualitative or b) quantitative
> Based on these results a risk profile is completed.
> In this profile the risks are compared with the
risk protection level, which determines the accepted
probability for defined consequences
> Use as an aid to prioritise actions!
EXAMPLE

 QRM for facilities, equipment and utilities
Assess an existing compressed air system
> Old approach: 60 “risks” should have been solved in detail
> Initial RM-Approach:
 4 sessions in total 16 people
 153 potential risks discussed
 34 Cases beyond the action limit
 30 Corrective actions have been performed (- 50%)
> Review of RM-Approach after inspection
 Did you consider this hazard?
- yes: show and explain rationale
- yes, but start discussion for a yes/no decision
- no: revisit initial risk assessment
EXAMPLE

Experiences
 Ease of applicability
> Prospective tool
> Good tool for operators to use
> Can be used to identify critical steps for validation
> More objective than Fault Tree Analysis
> Covers minor risks
EXAMPLE

Experiences
 Limitations
> Can be time and resource consuming
> Mitigation plans must be followed up
> Not a good tool for analysis of complex systems
> Compound failure effects cannot be analyzed
> Incorporating all possible factors requires a thorough
knowledge of characteristics and performance of the
different components of the system
> Successful completion requires
expertise, experience and good team skills
> Dealing with data redundancies can be difficult
Based on Takayoshi Matsumura, Esai Co
EXAMPLE

Annex I.3
Failure Mode,
Effects and Criticality
Analysis
(FMECA)

I.3: Failure Mode, Effects and Criticality Analysis (FMECA)
(IEC 60812)
 Extended to incorporate an investigation
of the degree of severity of the consequences,
their respective probabilities of occurrence and
their detectability
 The product or process specifications should be
established
 Identify places where additional preventive actions
may be necessary to minimize risks
Potential Areas of Use(s)
 Utilized on failures and risks associated with manufacturing
processes ICH Q9

Q9_Failure_Mode_Effects_A.ppt

More Related Content

Similar to Q9_Failure_Mode_Effects_A.ppt (20)

More from GeorgeVardas2 (7)

Recently uploaded (20)

Q9_Failure_Mode_Effects_A.ppt

Editor's Notes