Implementing a risk‐based CAPA process within a QMS is a necessity in the improvement of controls aligned with product and process non-conformances, adverse events, audit findings, complaints, etc. Making decisions concerning scope and extent about these “defectives” is a modern and cost-saving approach to improvement and compliance. Every non-conformity does not force you to open a corrective action. While almost every problematic issue needs at least a correction, the biggest payback is to use corrective actions on systemically-driven problems that are repetitive and recurring.
Applying the methods of determining risk to the device’s complete life cycle will give your company a complete look at all of the device’s risks—including those relative to processes. Manufacturers should be able to justify that they have reduced the risks as far as possible as part of their risk management plan and ongoing corrective and preventive actions.
Testing and Development Challenges for Complex Cyber-Physical Systems: Insigh...
CAPA: Using Risk-Based Decision-Making Toward Closure
1.
2. The Corrective & Preventive Action (CAPA) Process
“Why Is My CAPA Process Not Working?”
…..and I even asked “why” five times…..
John Gagliardi
MidWest Process Innovation, LLC
JGAGL777@One.Net
www.midwestprocessinnovation.com
3. What are the symptoms of an
ineffective CAPA system?
- __________________
- __________________
- __________________
- __________________
- __________________
- __________________
- __________________
Do these symptoms sound
familiar!
10. This is (also) the last slide of this session…..
1. Your root cause analysis is only as good as the info you
collect.
2. Your knowledge (or lack of it) can get in the way of a good
root cause analysis.
3. You have to understand what happened before you can
understand why it happened.
4. Interviews are NOT about asking questions.
5. You can’t solve all human performance problems with
discipline, training, and procedures.
6. Often, people can’t see effective corrective actions even if they
can find the root causes.
7. All investigations do NOT need to be created equal (but some
investigation steps can’t be skipped).
Y Y Y Y Y 9
11. CAPA FDA 483 Items
An informal review by the authors of recent FDA 483 Items issued
to medical device companies revealed that approximately 30%
of the cites were concerned with the firm’s CAPA system.
As evidenced by the significant number of warning letter cites on
the topic, firms face many challenges in making the CAPA system
work as intended.
12. Failure Investigation - Decisions
All non-conformities do not require a risk analysis, failure investigation,
and corrective action (every time)…
The preamble to the October 7, 1996 Quality System regulation provides insight
into FDA’s position on these matters. In comment 159 of the preamble which
relates to the degree of corrective or preventive actions, FDA states “FDA
cannot dictate in a regulation the degree of action that should be taken because
each circumstance will be different, but FDA does expect the manufacturer to
develop procedures for assessing the risk, the actions that need to be taken for
different levels of risk, and how to correct or prevent the problem from
recurring, depending on that risk assessment.” A CAPA system
should provide mechanisms to assess risk throughout the process.
16. Failure Investigation - Decisions
Nonconformities should be evaluated as to their potential risk to
patients and users. This evaluation should be supported by a
documented product risk assessment performed under design
controls or subsequently generated at the time the nonconformity is
identified. The determination whether or not to pursue failure
investigation and/or corrective action is based in part on the
magnitude of the problem and any related risk, and should be made
by properly trained and qualified individuals. The determination and,
if appropriate, the rationale for not pursuing corrective action
should be documented in accordance with criteria defined
in written procedures.
17. FDA’s August 1999 “Guide to Inspections of Quality Systems” (the QSIT manual) states,
“The purpose of the corrective and preventive action subsystem is to:
- collect information
- analyze information
- identify and investigate product and quality issues
- take appropriate and effective corrective and/or preventive action to prevent
recurrence
- verifying or validating corrective and preventive actions, communicating
- corrective and preventive action activities to responsible people, providing
- relevant information for management review, and documenting these
- activities are essential in dealing effectively with product and quality issues,
- preventing their recurrence, and preventing or minimizing device failures.”
18. CAPA Is a World-Wide Concept
The concept of CAPA is not restricted to the QS regulation or QSIT,
but is a widely accepted concept in most quality systems, whether
based on GMP, ISO, or some other standard or guideline.
Since most modern quality systems strive to improve quality, there
must be mechanisms in place to recognize existing or potential
quality issues, take the appropriate steps necessary to investigate
and resolve those issues, and, finally, make sure the same issues
do not recur.
CAPA process issues are basically the same world-wide
19. CAPA is Not Just About Product Defectives
The CAPA system must look beyond product issues and consider
other quality issues, including problems associated with systems
and processes.
During failure investigation or root cause analysis, firms will
occasionally conclude their investigation when the nonconforming part
or material is identified and corrected. However it is important to note
that an important element of the CAPA system is to address
processes and procedures.
In many instances, if a given process were working appropriately, it
would be less likely for nonconformities to be introduced into the
system. Where practical, processes and procedures should be
reviewed and considered as potential root causes for product
or quality issues.
20. Risk Management Plays a Major Role Concerning CAPA
RISK ANALYSIS
• Identification of
intended use,
purpose of
product/service
• Identification of
process hazards
• Estimation of risk
(occurrence/severity
and detection)
RISK
EVALUATION
• Decisions as to risk
acceptability
RISK CONTROL
• Option analysis
• Implementation
• Residual risk
evaluation
• Overall risk
acceptance
POST-
PRODUCTION
INFORMATION
Post-production
experience
Review of risk
management
experience
RISK MANAGEMENT
RISK ASSESSMENT
19
21. The Quality Review Board (QRB)
Overview
Cross-functional decision-making
Realization of Trends
Use of Quality Data
Team Approach
22. QRB - Risk Management Process
Documentation:
- Policy and Procedures toward Implementation
- Strong Linkage to Design Controls and Manufacturing
- Risk Management Files
- Hazards and Failure Modes Analysis
- Risk Controls
- Management Review
Correction Corrective Action Preventive Action
23. The Reach of the QRB
Risk Management System
Management
Responsibility
Audit Results
Complaints
N/C Products
Reports…corrections
& removals
Medical Device
Reports
Records, Documents
& Change Control
Quality Review Board
Medical Device
Tracking
Service Reports
Sterilization Process
Issues
Customer Feedback
22
24. Quality Review Board - Authority and Logistics
Responsibility and authority necessary for the QRB in
the Quality Management System…
- Cross Functional Team
- Pre and Post-market Inputs (as well as internal inputs)
- Risk-based Decision-Making
- Feedback to Management
- Feedback to Management Review
- Linkages to the CAPA Process
- Initialize Internal Investigations
- Trends in product and process performance
- Process effectiveness
- Product safety and functionality
- Supplier performance
- Success of performance improvement objectives
- Necessity for re-design
- Necessity for re-validation (process)
23
25. The CAPA Process Is Data Driven – Intro….. the QRB
The “Five Whys” should be information-driven
CAPA systems are (should be) inherently data driven. This means that
without adequate, relevant data, it can be difficult to draw definitive
conclusions about product or quality issues. There is an increasing
reliance on technology to help gather and sort data. One of the
challenges facing many firms is the proliferation of small, uncontrolled
data repositories within the organization. Departments or individuals
now have the ability to create and manage databases that house
important quality information that is invisible to the rest of the
organization. Where practical, firms should seek out and use tools that
allow for the collection of that information in a more centralized
manner, while still allowing those same departments and individuals to
do their respective jobs.
26. An important aspect of the CAPA system is that the data coming into
the system are complete, accurate and timely. Your QRB should
implement procedures that describe:
- What the sources of product and quality data are
- How the data are to be captured
- How the data is to be analyzed, including the method of analysis
- When the data is to be analyzed
- What steps are taken after analysis
QRB should also be able to demonstrate that the activities described
above actually occurred.
Data that is Complete, Accurate and Timely
…..toward understanding root cause…..
27. Input into the Management Review
Output from Management Review
Visibility
Evidence to show an FDA investigator that CAPA information is an integral part of the
management review process might include:
- Procedures for management reviews that contain standard agendas with headings
that discuss the review of corrective and preventive actions.
- Agendas relating to the management review of the quality system including
corrective and preventive actions, and certification that reviews were conducted.
- Completed or closed corrective and preventive actions that resulted from the
management reviews.
- Schedules of management reviews (past and future)
- CAPA analysis reports submitted for Management Review.
- Top – level visibility
29. Sources of Quality Data – to make decisions
- Acceptance activity records relating to component, in process, and device testing
- Complaints
- Medical Device Reports (MDRs) and Vigilance Reports
- FDA 483s and Warning Letters
- Reports of system, process or product nonconformities
- Process monitoring data (e.g., statistical control charts, trends, run charts, etc.)
- Calibration and maintenance records
- Scrap, rework and “Use As Is” (UAI) records
- Clinical adverse events
- Internal, external, supplier and third party audits
- Returned products analysis
- Installation and/or repair reports
- Spare parts usage
- Customer and/or technical service requests
- Field service and/or warranty reports
- Customer feedback (e.g., surveys, polls, etc.)
- Historical records from previous corrections
- Corrective and preventive actions
- Lawsuits and other legal actions
- Published literature
- Reports from employees www.midwestprocessinnovation.com
intelligent
Y Y Y Y Y
30. Statistical methods can also be useful when they are applied across
different data sources to analyze comparable issues.
For example, by comparing manufacturing Non-conformances across
multiple manufacturing sites, information regarding common
suppliers, or complaint information against customer service reports,
the firm can achieve a more complete perspective of the entire
organization. Although statistical techniques are powerful tools for
analyzing product and quality issues, they should not be the only
mechanism used by firms.
29
Trending is an Important Statistical Tool
31. Non-statistical techniques provide an opportunity to rely on corporate
knowledge and expertise to recognize issues sometimes before there
is an observable trend. These techniques include (in part):
- Management reviews
- Quality and/or material review boards
- Safety committees (internal or external)
- CAPA Disposition Meetings
- Employee suggestion programs
- Individual clinical, product or engineering expertise
- Other internal reviews
30
Non-statistical Techniques…..also an important tool
33. 32
5.1 Measure and 5.2 Analyse
coordination / linkage of data / data sources / “horizontal analysis”
5.0 Phase II
Measurement and
Analysis within and
across Data Sources
6.0 Phase III
Improvement
4.0 Phase I
Planning
4.1 Plan for Measurement, Analysis and Improvement Processes
7.0 Phase IV
Input to Management 7.1 Report to Management and 7.2 Management Review
Examples of defined Data Sources
Within each data source
4.2 Establish Data Sources and Criteria
Feedback
Complaints
ServiceReports
Sparepartsusage
QualityAudits
(internal/external)
Supplier
Performance/Controls
ProcessControls
ReturnedProduct
Market/customer
survey
Improvement
6.1 Investigate 6.2 Identify Root Cause
6.3 Identify Actions
6.4 Verify
identified Actions6.5 Implement Actions
6.6 Determine
Effectiveness of
Implemented Actions
34. A compliant CAPA system “closes the loop” on many of the documented
issues by directly providing input into the design control requirements of its
quality system.
For example, trending is realized from non-conforming product reports and
this information is relayed to the QRB for purposes of decision-making. The
risks are evaluated, the product non-conformity is investigated and the QRB
decides that this is a design issue and not merely a manufacturing non-
conformance.
33
Closed Loop Process
35. (part of) Failure Investigation
…..and you ask Why?.....
FIVE TIMES
34
36. SIMPLE TECHNIQUE: 5-Whys
I’ve probably heard more “experts” talk about 5-Whys than any other
root cause tool. Why? Because it is simple. Simple to teach and simple
to use. All you have to do to find root causes is ask “Why?” five times.
Here’s an example of the technique from the technique’s creator, Tailchi
Ohno:
1. “Why did the robot stop?”
The circuit has overloaded, causing a fuse to blow.
2. “Why is the circuit overloaded?”
There was insufficient lubrication on the bearings, so they locked up.
3. “Why was there insufficient lubrication on the bearings?”
The oil pump on the robot is not circulating sufficient oil.
4. “Why is the pump not circulating sufficient oil?”
The pump intake is clogged with metal shavings.
5. “Why is the intake clogged with metal shavings?”
Because there is no filter on the pump.
35
37. What do you think? Is “NO FILTER ON THE PUMP” a root cause? I
think this example is a perfect example of what is WRONG with
5-Whys.
First, they missed a whole line of questioning. Why didn’t the loss of
lube oil pressure trigger an alarm or an automatic shutdown?
Another line of questioning that was missed was “Where did the
metal shavings come from?” After all, metal shavings are not
normally found in a well-maintained machine.
And finally, ”Why was there no filter on the pump?” Did maintenance
forget to install it? Did the designer fail to include it? Was it removed
because it kept getting clogged?
All of these questions need to be answered but the ultimate expert,
Tailchi Ohno, didn’t ask them because he thought he already
had the answer.
36
39. Inexperienced investigators…..They can’t find what they don’t know.
Many “experts” have little or no training or understanding of the psychology behind human error.
They search for familiar patterns and disregard counter evidence. (The technical name for this phenomenon
is “confirmation bias.”)
But many investigators, and some root cause tools, start by asking “Why?” when they should be trying to
understand “What happened?”
Starting by asking “Why” is jumping to conclusions. And this can lead the investigator to find causes that
they have jumped to because they didn’t first seek to understand.
The “Why” question turns off the “remembering” trail that we want the brain to go down and turns on the
“justification” trail. After all, isn’t the purpose of an interview to collect information (not justification)?
If you look at most industrial accident/incident investigations, you find three standard corrective actions:
1. Discipline. Which starts with the common corrective action: “Counsel the employee to be more careful when …”.
2. Training. This may be the most used (and misused) corrective action of all.
3. Procedures. If you don’t have one, write one. If you already have one, make it longer.
The misuse of these three standard corrective actions is the reason that so many accident investigations don’t really
cause performance to improve. They don’t solve the real problems.
38
40. Failure investigation procedures might include provisions for:
- Establishing who is responsible for evaluating product or quality issues and
determining if a failure investigation is necessary.
- Maintaining a record when no failure investigation is made, including the
reason and the name of the individual responsible for the decision.
- Describing how investigations are conducted and what records are
maintained.
- Defining the content and format of the failure investigation report.
- Identifying the failure modes…..FMEA
- Determining the significance and risk of each failure mode.
- Determining the depth to which a failure investigation is to be carried out
including when an investigation should include root cause analysis.
- Requiring review and approval of the failure investigation, risk analysis,
conclusions and any required actions.
- Feeding information into FMEA (Failure Effects and Modes Analysis) and
tying back to the risk analysis originally performed during the design phase.
39
41. Often, people can’t see effective corrective actions even if they can find the
root causes. Why? Because they have performed the work the same way for
so long that they can’t imagine another way to do it.
It’s true that some investigations may take too long and cost too much. But
that doesn’t mean that every root cause analysis needs to take too long and
cost too much. Root cause analysis should be scaled to the size of the
problem and the risk of future accidents with similar causes. Small risk =
small investigation. Big risk= ___________
40
42. 1.Your root cause analysis is only as good as the info you collect.
2. Your knowledge (or lack of it) can get in the way of a good
root cause analysis.
3. You have to understand what happened before you can
understand why it happened.
4. Interviews are NOT about asking questions.
5. You can’t solve all human performance problems with
discipline, training, and procedures.
6. Often, people can’t see effective corrective actions even if they
can find the root causes.
7. All investigations do NOT need to be created equal (but some
investigation steps can’t be skipped).
41
43. Failure Investigation – Basic Expectations
The comprehensiveness and depth of failure investigations should be
commensurate with the magnitude of the issue, and the potential risk the
issue presents to the patient and/or end user. Typically, the failure
investigation should include:
- Problem identification and definition – Once identified, the issue must be
characterized and defined in order to understand the potential scope and
impact. This characterization should include the documented risk analysis.
- Investigation – The issue is investigated and documented, including results
of analyses (data, process, operations and/or other sources of information)
and identification of failure modes and conclusions regarding root cause(s).
- Rationale – If a failure investigation or analysis is not performed, the record
should include the reason no investigation was performed and the
individual responsible for the decision.
- Action Plan – A thorough failure investigation might include information and
other data that can be useful in crafting an action plan to address the
issue.
44. Significance of the Issue – Risk-related rationale
Typically, the following factors are taken into consideration in determining if a product
or quality issue is significant:
- Is there potential for user or patient safety issue? – Likelihood that the issue may
result in a death or serious injury, even if the cause is determined to be user error.
- What is the classification of the device? – Class II and III products tend to have
higher risks associated with them than Class I products.
- Is there a reliability issue? – Does the nonconformance or product issue affect
the product’s reliability?
- Did the product meet its specifications? – Does the nonconformance or product
issue cause the product to fall outside of established specifications?
- Was the product labeling involved? – Does the nonconformance or product issue
cause the product to be misbranded, adulterated, or otherwise not properly identified?
- Has the frequency of occurrence for a known issue changed? – Is the
nonconformance or product issue occurring at a higher rate than expected over a
given period?
- Is the issue difficult to detect? – A nonconformance or product issue that is
difficult to detect is less likely to be identified and corrected prior to causing a
problem.
46. Evidence to show an FDA investigator that the CAPA
system is functioning properly
Evidence to show an FDA investigator that the CAPA system is functioning
properly might include:
- CAPA inputs – Sources of product and quality data are routinely reviewed.
- Analysis – Documentation demonstrating that data were analyzed for
possible unfavorable trends or other indicators of product or quality issues.
- Investigations – Documentation demonstrating that qualified personnel
have investigated unfavorable trends, non-conformances, and product
issues.
- Action records – Corrective or preventive action plans and other records
demonstrating action start and end dates.
- Effectiveness checks – “Before” and “After” product or process quality trend
reports or other analyses. These checks should establish that the action
effectively prevented the recurrence (or initial occurrence) of the
non-conformance.
47. Evidence to show an FDA investigator that the CAPA
system is functioning properly
Evidence to show an FDA investigator that the CAPA system is functioning
properly might include:
- Verification checks – Engineering change orders (ECOs) or other control forms
for initiating a change to a product or process.
- Validation checks – Design verification, design validation, and process
validation protocols, and summary reports to ensure the actions were
effective and that they do not adversely affect the device.
- Control checks – Training records, product tracking information, segregation,
reprocessing, rework or final product disposition records with appropriate
authorizations.
- Timeliness checks – Tracking mechanisms to demonstrate that actions are
conducted in a timely manner.
- Management checks – Management review agendas and schedules,
corrective/preventive action status, aging and closeout reports.
48. Remember…..
1. Your root cause analysis is only as good as the info you collect.
2. Your knowledge (or lack of it) can get in the way of a good root cause analysis.
3. You have to understand what happened before you can understand why it
happened.
4. Interviews are NOT about asking questions.
5. You can’t solve all human performance problems with discipline, training, and
procedures.
6. Often, people can’t see effective corrective actions even if they can find the
root causes.
7. All investigations do NOT need to be created equal (but some investigation
steps can’t be skipped).
Y Y Y Y Y
47
