Panel Discussion

FDA and Industry Collaboration on Computer Software Assurance (CSA)

20th Annual Computer And IT Systems Validation, April 23, 2019

Panelists:
➢ Khaled Moussally, Compliance Group – Global Head of QMS
➢ Jason Spiegler, Siemens Digital Industries Software – Senior Director,
Strategic Initiatives & Customers, Life Sciences Practice
➢ Kurt Weber, Vericel Corporation - Director of Information Technology
➢ Harsha Chulki, ICU Medical - Head of Global IT Quality & CSV
➢ Ken Shitamoto, Gilead Sciences – Sr Director, IT

Material Contributors:
➢ Cisco Vicenty, Office of Compliance, FDA (CDRH)
1
Objectives:
➢ Create awareness to accelerate innovation
➢ Inspire action so you can begin to realize value

Agenda:
➢ Quick recap on FDA CSV Team’s journey and CSA Draft Guidance
➢ Panel discussion on:
o Recent recommendations to the Draft Guidance
o Success Stories: Vericel, Gilead Sciences and ICU Medical
o “Test More; Document Less” “Computer Software Assurance for
Manufacturing, Operations, and
➢ Q&A and Open Discussion
Quality System Software”
Computer System Validation!

What do you love about it?

What do you dislike about it?

What’s the background?
Journey of FDA CSV Team
FDA Case for
Siemens – Fresenius
Quality begins Industry team formed /
2011 - Executive Exchange w/ FDA:
2015 recommendation development begins
2012 CSV Barrier identified
Q2
2016
Q1

Begin promoting recommendations:

2017
Zoll Lifevest + Medtronic value examples

• More industry adoption

• FDA “A List” status for CSA • CSA Draft Guidance
Draft Guidance release targeted for 2019
2018
• More examples developed
• More firms applying
2019
recommendations (Vericel,
ICU Medical, Gilead, etc)

www.fda.gov
 CSV identified as a barrier for the FDA…

For your technology

investments, what are the
barriers for Realizing Value?

CSV!!!

www.fda.gov
 The Industry CSV Team

Company Name Company Name

Baxter Healthcare Tina Koepke Johnson and Johnson Dana Guarnaccia
Boston Scientific Damien McPhillips Johnson and Johnson Ron Schardong
Boston Scientific Ray Murphy Lantheus Imaging Lou Poirier
Compliance Group Khaled Moussally Medtronic Frankie Bill
Edwards Lifesciences Penny Sangkhavichith Medtronic Michael Branch
Edwards Lifesciences Andy Lee Medtronic April Francis
FDA Cisco Vicenty NeuroVision Imaging Pepe Davis
FDA John Murray Ortho-Clinical Diagnostics Des Chesterfield
Fresenius Medical Care Bill D'Innocenzo Siemens PLM Jason Spiegler
Fresenius Medical Care Curt Curtis Siemens PLM Greg Robino
Fresenius Medical Care Marc Koetter Siemens PLM Thorsten Ruehl
Zoll Lifevest Frank Meledandri Sr.
Contributions also provided by past team members:
Stacey Allen, Jason Aurich, Sean Benedik, Laura Clayton, Bill Hargrave, Joe Hens , Scott Moeller & Mark Willis 8
 Non-Product CSV is foundational …

for enabling the “Digital Thread” of Smart Manufacturing

8
www.fda.gov
What are the pain points in CSV?
Common CSV Pain points

• Takes forever

• Too much focus on document generation

• Perceived Regulatory burden - doing things for the auditor

• Test Scripts that often run into tens of pages

• Test Script errors & costly defect management

• Paper based processes

• Complex, hard to use Risk-based approaches

Quick recap of FDA – Industry CSV Team’s
recommendations and what’s new?
Key Take Aways
Key
Why Now?Take Aways
Med Dev lags other industries
• Lack of clarity
• Outdated compliance approach 80%
• Perceived regulatory burden 20%
• Reduces manufacturer’s Test Document
capability to learn, react, &
% Time Spent
improve

www.fda.gov
Key Take Aways
Key
Why Take Aways
Now? Defining Risk
• Clearly define “intended use”.
Create a Paradigm Shift…
• Focus on the “direct impact on device safety and
• Streamline with value-driven, device quality”, and does it result in “patient/user
patient focused approaches 80% safety risk?” See examples.
• Critical thinking & risk-based 20% ➢ LMS vs Manufacturing Equipment Software
agile approaches • For PMA Products, CDRH is exploring using risk
Test Document
• Improve manufacturer’s determination to make implementation of systems an
% Time Spent annually reportable change no 30-Day Notice
capabilities with automation

Risk Based Assurance Strategies Assurance Evidence Capture

• Take credit for work already done • Use CSV tools to automate assurance activities
➢ Leverage existing activities and trusted supplier data Note: FDA does not intend to review validation of
• Use Agile test methods (e.g. unscripted testing) support tools.
when appropriate
• Use electronic data capture &
• Mitigate risk with downstream process controls record creation vs paper
• Leverage continuous data and information for documentation, screen shots,
monitoring and assurance etc.
www.fda.gov
 Acceptable record of results
Assurance Approach Test Plan Test Results Record (Digital Acceptable)
Unscripted Testing: • Testing of features and functions • Details regarding any • Summary description of features and functions tested
Ad-hoc (with least- with no test plan failures/deviations found • Issues found and disposition
burdensome • Conclusion statement
documentation) • Record of who performed testing and date
Unscripted Testing: • Testing of feature and function fail- • Details regarding any • Summary description of fail-modes tested
Error guessing modes with no test plan failures/deviations found • Issues found and disposition
• Conclusion statement
• Record of who performed testing and date
Unscripted Testing: • Establish high level test plan • Pass/fail for each test plan • Summary description of features and functions tested
Exploratory Testing objectives for features and functions objective • Result for each test plan objective – only indication of pass/fail
(no step-by-step procedure is • Details regarding any • Issues found and disposition
necessary) failures/deviations found • Conclusion statement
• Record of who performed testing and date
Scripted Testing: • Limited Test cases (step-by-step • Pass/fail for test case identified • Summary description of features and functions tested
Limited procedure) identified • Details regarding any • Result for each test case - only indication of pass/fail
• Expected results for the test cases failures/deviations found and • Issues found and disposition
• Identify unscripted testing applied disposition regarding fails • Conclusion statement
• Independent review and approval of • Record of who performed testing and date
test plan. • Signature and date of appropriate signatory authority
Scripted Testing: • Test objectives • Pass/fail for test case • Detailed report of assurance activity
Robust • Test cases (step-by-step procedure) • Details regarding any • Result for each test case - only indication of pass/fail
• Expected results failures/deviations found and • Issues found and disposition
• Independent review and approval of disposition regarding fails • Conclusion statement
test cases. • Record of who performed testing and date
• Signature and date of appropriate signatory authority
18
www.fda.gov
 Spreadsheet to Analyze and Graph Non-conformances
The manufacturer developed a spreadsheet used to analyze, and graph non-conformances stored in
a controlled system. Intended use of the spreadsheet is found to have a low patient risk.
• Intended Use: Analyze and graph non-conformances data stored in a controlled system

• Risk Assessment: The intended use of the spreadsheet is for analyzing process quality outcomes and is identified as a
high-risk function. The manufacturing process includes additional changes and inspections that assure non-
conformances do not escape therefore the patient risk is low.

• Tested: Spreadsheet X, Version 1.2

• Test type: Unscripted testing – exploratory testing

• Goal: Ensure that analyses can be Created/Read/Updated/Deleted

• When/Who: July 9, 2019, by John Smith

• Testing activities: Created, updated, and deleted analyses and observed that all calculated fields were correctly updated

• Conclusion: No errors observed

That’s it!
No more documentation required!
15
www.fda.gov
 Complaint Handling Spreadsheet example
Patient/Quality System Risk Examples
Errors in the spreadsheet outputs • Manufacturing equipment controls Implementation Method
will cause death, life-threatening
• Automated product inspection Out of The Box Configured Custom
harm or irreversible damage to
• Label management & automation High Limited Scripted Robust Scripted

Patient/Quali
High patient and / or fail the direct Unscripted Testing
• QC Laboratory test calculations Testing Testing

ty System
implementation of a quality Medium Ad-Hoc Testing (With Limited Scripted
system activity defined in a • Adverse event tracking Record)
Unscripted Testing
Testing
•

Risk
regulation Clinical trial results Low Ad-Hoc Testing (With Ad-Hoc Testing
Unscripted Testing
Errors in the spreadsheet outputs • Product Complaint Tracking Record) (With Record)
will cause significant but • Product quality status management
temporary harm or reversible • Product Complaints Trending
Medium damage to patient and / or fail to • Spreadsheet supporting managers review Spreadsheet Description / Intended Use Patient/Quality Implementation Assurance Activity
support the indirect the training assigned to either to their Risk Level Method
implementation of a quality employees (the system that directly Product Recall Decision Matrix – Assessment High Custom Robust Scripted
system activity defined in a implements the quality system activity is form used to decide the “recall” status of a Testing
regulation. the Learning Management System product based on predefined trending criteria
Errors in the spreadsheet outputs Complaint Trending Tool – Extract complaint Medium Custom (VBA Limited scripted testing
will cause minor harm to a data from various sources and identify complaint Macros)
• Company specific quality trending
Low patient and / or fail to implement trends
spreadsheets not required by regulation.
a quality system activity not Non-conformance data analysis - Analyze and Low Configured Ad-hoc testing
defined in a regulation graph non-conformances data stored in a
• Manufacturing cost reports controlled system
• Turnaround time reports
• Project Management tools
No direct or indirect impact on
None • Task tracking, Schedule Implementation Method Examples for Spreadsheets
patient
• Activity Due Date Calculator (Non-
Printing functions, Formatting functions, Simple Arithmetic Operations (e.g. sum, average,
Product) Out of the Box
std.dev), Simple Cell Relationships and Range Operations
• Vendor Assessment Checklist
Formulae (including statistical functions), Single level Boolean functions (If, And, Or),
Configured
Charts, Conditional Statements, Nested Functions, Security (Cell Protection), Pivot tables
Custom Macros, VBA Code, Extensive Nested Logic Functions (e.g. IF(condition1,
Custom value_if_true1, IF( condition2, value_if_true2, IF(condition3, value_if_true3,
value_if_false3)))
Qualification of Automated CSV Tools

These intended uses of these tools are not part of production or the quality system, therefore should
not be considered within the requirements of 21 CFR 820.70(i).

Examples include:
• Code Debugger (for CAPA Automation use case)
• Loadrunner (for simulating anticipated peak load of ERP production system use case)
• Defect management and ALM tools

Assurance Process
Step 1: Identify where and how tool will be used within your organization
Step 2: Determine if the Off the Shelf tool is part of, integrated, or used for automating production or Quality Management
Step 3: Assure the use within your organization
Step 4: Capture Evidence
Can you provide examples of applying these
recommendations and the resulting value?
Vericel Industry Case Study – Unscripted Testing

Standard FDA – Industry

Scripted Testing Streamlined
Approach Approach

80% 80%

20% 20%
LEGACY ERP, EBR
Test Document Test Document
SYSTEMS Systems
% Time Spent
% Time Spent

Production Scheduling system (much ERP System validated in 3 months

smaller than ERP) took much longer
Approach leveraged for EBR system
80% issues were test script issues
90% reduction in test script issues
Focus on documentation, not on testing
> 50% reduction in validation spend &
High validation spend & time time
Vericel Industry Case Study – Risk Framework
Legacy Risk Assessment New FDA CSV Risk
Framework Framework

80% 80%

20% 20%
Focus on Focus on Focus on Focus on
Testing Documentation Testing Documentation
% Time Spent % Time Spent

Legacy Risk Framework FDA CSV Risk

Worksheet Framework Worksheet

Req/Spec ID(s) Patient Risk Mitigation Strategy Test Strategy

Risk Rating

5 Required Mitigation 1 Scripted Testing

Spec ID #1 High

Unscripted Testing –
Spec ID #2 Medium 3 Required Mitigation 2
Error Guessing
Unscripted Testing –
Spec ID #3 Low 2 N/A
Exploratory Testing
ICU Industry Case Study – Unscripted Testing

Legacy Approach FDA CSV Team Approach

Time for drafting and Reduction in validation

2X 75%
executing testing cycle times

More Test Script & Tester Reduction in Test Script &

95% errors
95% Tester errors

Reduction in Validation
50% Less system issues found 60% Testing spend
A COMPARISON
FDA CSV Team Practice
Vs
Legacy Practice
Gilead Empirical Retrospective Validation Effort Comparison of Traditional vs. New Model
Analysis
System Name
80%

70% Change Management System*

(Jan 2014)
60%
Electronic Document Management System
(Dec 2014)
50%

Laboratory Execution System

40% (May 2015)

30%
Process Control System*
(May 2016)
20%
Environmental Monitoring System
(Jul 2016)
10%

Product Serialization System

0% (Aug 2018)
CMS EDMS LES PCS EMS PSS

SAVINGS DEVIATION SLIPPAGE MODEL SLIPPAGE * Bold = favorable audit

 How have these Recommendations
been received by the auditors/investigators?
CSV Industry Partner “Quote”
“………….sitting here with “EU Notified Body” auditor for our 13485-2016 audit.. just
talked software application validation for the last hour. Our auditor was impressed
with everything we have set up, loved the risk based approach. Told us that other
companies get into lots of arguments with him over this and it was refreshing to talk
with us. He personally likes defined OQ verbiage, however how I explained we do our
test cases in test and live environments and he agreed it was just a wording thing on
his part. We also talked a lot about the difference between verification activities vs
validation activities and effectively setting up test cases. He was impressed with
everything we had………………”

“…………………….It was awesome finally going through a validation and being assured
that things were good from the EU perspective as well...
- Frank M.
 What are your next steps?
How are you planning to automate?
 Next Steps
• Continue developing Use Cases and new recommendations
• Encourage manufacturers to start using recommendations
❑ Capture value - measure better, faster, and less expensive CSV activity and…

❑ Provide FDA with input on what is working vs not working, barriers, etc

• FDA 2019 “A List” - new Non-Product CSV FDA Guidance to encourageautomation

and clarify expectations for risk-based CSV.
Your assignment: Provide comments to the Docket!

• Review and possibly modify 820.70(i) – improve clarity on automated processes

• Promoting recommendations through recorded Webinar:

https://www.plm.automation.siemens.com/country/en-us/webinar/fda-non-product-csv/29180/index.html

32
 For Questions

Contact:
Khaled Moussally ([email protected])
Jason Spiegler ([email protected])
Cisco Vicenty ([email protected])

33
BONUS MATERIAL
 FDA’s View of Automation
The FDA supports and encourages the use of automation, information technology, and data solutions throughout the
product lifecycle in the design, manufacturing, service, and support of medical devices. Automated systems provide
manufacturers advantages for reducing or eliminating errors, increasing business value, optimizing resources, and
reducing patient risk. Is based on learning from other industries where automation has already shown significant
benefits in enhancing product quality and safety, which in turn reduces Risk, compared with non-automation.
www.fda.gov
 FDA CDRH Fiscal Year 2019 (FY 2019)
Proposed Guidance Development
Draft Guidance Topics
❑ Content of Premarket Submissions for Cybersecurity of Medical Devices of Moderate and Major Level of Concern
❑ Surgical Staplers and Staples – Labeling Recommendations
❑ Non-binding Feedback After Certain FDA Inspections of Device Establishments
❑ Select Updates for Recommendations for Clinical Laboratory Improvement Amendments of 1988 (CLIA) Waiver Applications for Manufacturers of In Vitro Diagnostic Devices
❑ Recommendations for Dual 510(k) and Clinical Laboratory Improvement Amendments Waiver by Application Studies

❑ Computer Software Assurance for Manufacturing, Operations, and Quality System Software
❑ Patient Engagement in Clinical Trials
❑ Guidance for the Content of Premarket Submissions for Software Contained in Medical Devices
❑ Lifecycle Regulatory Requirements of Medical Device Servicing (Device Servicer vs Remanufacturer)
❑ Guidance on an Accreditation Scheme for Conformity Assessment of Medical Devices to FDA-Recognized Consensus Standards (ASCA).

FDA announcement link. https://www.fda.gov/MedicalDevices/DeviceRegulationandGuidance/GuidanceDocuments/ucm529396.htm

9
www.fda.gov
 Scope of the Guidance

Scope
• Manufactures, including combination product manufactures, who have established a 21 CFR 820 quality system.

Out of Scope
• Automation of Clinical Trial and associated Electronic Records
• Automation regulated outside of 21 CFR 820
• Software validation for devices under the requirements of 21 CFR 820.30(g).
• Information Technology Infrastructure assurance activities such as planning, contracting, or continuity of operations
efforts that are used to maintain server operations.

9
www.fda.gov
Focus on Assurance Shift the discussion

www.fda.gov
 (Unique) Clarifications and Recommendations

Intended Use Risk Based Approaches

• What is the intended use? • Do automation features, operations, or functions directly impact device
• Does feature, operation, or function directly impact safety or device quality?
➢ device safety ➢ High-risk areas may require the most rigorous assurance effort to
ensure they perform as intended.
➢ device quality or
• FDA intends focus on areas that Directly impact device safety or device
➢ quality system integrity? quality. FDA does not intend to focus on Indirect impact areas. Ex: MES or
LIMS compared with an LMS.

Assurance (Testing) Approaches Evidence Capture Methods

• Provide confidence that the system, feature, or • Least-burdensome record (see next slides). Record needs to be of value to
function performs as expected and meets intended use. the Manufacturer, not the Investigator or Auditor.
• Assurance Activities driven by the Risk associated with • CSV tools encouraged to automate assurance activity. Use electronic data
the system, feature, or function, depending on how you capture and record creation (vs paper documentation, screen shots, etc).
approach it (e.g. Direct vs Indirect). ➢ 21 CFR 820.70(i) is applied only when software part of production or
• Traditional IQ/OQ/PQ is not necessary for CSV. quality system. FDA does Not Intend to review validation of support
• Next slides will include examples of assurance activities, tools. Manufacturer responsible for determining assurance.
including numerous Agile testing methods. ➢ Part 11 narrowly scoped & under enforcement discretion (apply
appropriately)
33
www.fda.gov
 What does FDA care about? Risk
Considerations
• Direct impact to device quality and device safety that
also has a direct patient safety risk
• Directly impacts physical properties of the product or
manufacturing process identified as essential to device
safety or device quality by the manufacturer
• Measures, inspects, analyzes, and or dispositions the
product or process
• Determines acceptability or performs process corrections
without human intervention, awareness, or review
• Directly impacts labeling, instructions for use, or direct
alerts or communications to the user
• Automates surveillance, trending, or tracking of product
quality or patient safety issues identified as essential by the
manufacturer

16
www.fda.gov
 Appropriate methods and activities for
software assurance
• Take a least-burdensome approach – focus on value for the Manufacturer, not
the Investigator.
• Leverage existing activities and supplier data. Do not reinvent the wheel; take
credit for work already done
• Leverage use of process controls to mitigate risk
• Use Computer System Validation tools to automate assurance activities
➢ Scope of 21 CFR 820.70(i) is applied when computers or automated data processing
systems are used as part of production or quality system.
➢ FDA does not intend to review validation of support tools. Manufacturer
determines assurance activity of these tools for their intended use.
➢ Part 11 narrowly scoped and is under enforcement discretion apply appropriately
• Use Agile testing methods and unscripted testing as appropriate
• Use electronic data capture and record creation, as opposed to paper
documentation, screen shots, etc
• Leverage continuous data and information for monitoring and assurance

17
www.fda.gov
 Examples

Restricted © Siemens AG 2018

Page 19 2018.10.19 Siemens PLM Software
FDA CSV Team Recommendations – Case Study Themes

Embrace
Automation

Implement
Embrace Consistent Risk
Automation
Implement Adopt patient Follow
Scale Risk-based
the level Assurance
oftodocumentation
Adopt Patient-centric
•• Automation
Frameworks needs approaches
be seen
consistent risk centric •Scale level of testing based
Scale the level ofand
SDLConas an enabler.
patient-centric
documentation, not just
•• Put
Adopt
risk
patient front
doesn’t
a consistent
Automation
rating.
center
risk
need to of
frameworkall decisions.
involve across
complex
frameworks approaches •• testing,
FDA
SDLC,
tools.CSV based
Teamon
Change
risk rating. a patient centric
recommends
Management & Periodic Review
• Focus on
Leverage to testing,
vendor not on scripting.
documentation.
approach
Cycle. validation.

Scale the level Follow

of Risk-based
documentation Assurance
Non Product CSV Success Story: Value of FDA Collaboration

Scope:
• Camstar MES
• Epicor ERP

Best practices leveraged from

the Industry Team:
• Vendor Qualification
• Unscripted / Ad-Hoc Testing
• Analytics and Reporting
• Automated CSV Risk
Management

Presented by Frank Meledandri Sr

Siemens Software – Medtronic – FDA Design Excellence Event, May 15, 2018
www.fda.gov
*Presented by Medtronic on March 12, 2019 at IVT Medical Device Week in Minneapolis MN
*Presented by Medtronic on March 12, 2019 at IVT Medical Device Week in Minneapolis MN
*Presented by Medtronic on March 12, 2019 at IVT Medical Device Week in Minneapolis MN
*Presented by Medtronic on March 12, 2019 at IVT Medical Device Week in Minneapolis MN
 Case Study Examples – Embrace Automation – Infrastructure Qualification

FDA CSV Team • Use electronic data capture and record creation, vs paper documentation, screen shots, etc.
Recommendation • Leverage continuous data and information for monitoring and assurance

• Replaced manual, paper based test evidence capture with an automatedapproach.

Success Story
• Replaced manual, error-prone specification maintenance with an automated, error-free
Brief Description specification generation approach.

Before After

• Manual screen shots of evidence of server’shardware • Automated reports of server’s hardware and software
and software specifications. specifications by installing monitoring tools on servers.

• Manual and reactive maintenance of infrastructure • Automated, proactive generation of infrastructure

specifications – specifications are often not in sync specifications with the click of a button. Continuous data
with the actual infrastructure as infrastructure is so monitoring and assurance.
dynamic.

• Time taken – 10X • Time taken – 1X

 Case Study Examples – Embrace Automation – Smart Glasses

• Use electronic data capture and record creation, as opposed to paper documentation
FDA CSV Team • Use Computer System Validation tools to automate assurance activities
Recommendation • FDA does not intend to review validation of support tools. Manufacturer determines assurance activity of these
tools for their intended use.
• Replaced travel-intensive, hands-on training with remote, hands-free training using Smart Glasses
Success Story
(A wearable, voice-recognition & AI based technology)
Brief Description • Automatic, hands-free, safe evidence capture & voice-enabled real time, online documentation
Before After
• In person training (with expensive travel) required per • Remote training using wearable, hands-free, AI powered
procedures in order to perform certain manufacturing Smart Glasses technology.
tasks.

• Hands-on picture capture with external camera, print • Hands free evidence capture with voice-powered real-
out and attach to documentation offline. Error prone. time documentation. Error free.

• Deviations due to missed output recordings. • No deviations due to missed recordings.

• Time taken – 5X • Time taken – 1X

 Case Study Examples – Risk based Assurance – Consistent Frameworks

• FDA is interested in the situations when a failure to fulfill the intended use of the system, software, or
FDA CSV Team
feature directly impacting device safety and device quality results in direct patient safety risk.
Recommendation

• Deployed a patient centric risk framework across Software life-cycle – i.e. Validation, Change
Success Story
Management & Periodic Reviews.
Brief Description • Leveraged FDA CSV Team’s risk assurance framework.
Before After
• Siloed risk frameworks across processes – • Consistent, simplified risk framework across processes that
frameworks that don’t talk to each other drive a common risk based assurance approach

• Confusion among implementing teams with risk • Consistent implementation of harmonized risk
definitions that don’t align with each other assurance framework

• Redundant work efforts due to misalignment • Reduced cycle times from consistent interpretations
across processes
 Case Study Examples – Risk based Assurance – Deliverable Scalability

• FDA is interested in the situations when a failure to fulfill the intended use of the system, software, or
FDA CSV Team
feature directly impacting device safety and device quality results in direct patient safety risk.
Recommendation

• Deployed a software validation framework in which deliverables are scaled based on risk level of the
Success Story
software.
Brief Description • Leveraged FDA CSV Team’s risk assurance framework.
Before After
• One-size-fits-all list of validation documentation for • Deliverables scaled (both quantity & quality) by using a risk
all types of software assurance framework included in FDA CSV Team’s
recommendations

• Creation of documentation – not assurance • Creation of “assurance” – not just documentation

• Time consuming validation cycles • At least 25% improvement in validation cycles

 MES Example
Implementation Definitions
Feature works simply by installing the software and
Out of the Box adding necessary master data (e.g. products, BOM,
routes, etc.).

Feature is enabled through the setting of parameters

Configured
without changing the code of the software
Feature requires programming or change to software
Custom
code

Replacing custom MES with COTS MES.

Vendor Qualification: Mature vendor, trusted in industry, ISO certified, very
transparent, robust SDLC, etc.
Assurance Activity
Feature Patient Risk Level
Custom MES *COTS MES

Training Low 1 - Vendor

(Product quality inspected 3 - Unscripted
Enforcement at multiple steps in process)
Audit

Material Expiration 4 - Limited

Medium 2 - Ad Hoc
Enforcement Scripted
Label Printing High 5 - Scripted 3 - Unscripted
47
www.fda.gov *Out of the Box implementation
 Example: Annual Filing vs 30 Day PMA Notification
Scenario 1: COTS MES to replace paper based manufacturing / DHR
• Intended Use: MES will automate functions currently performed manually, providing 5M enforcement, data collection,
track/trace, label printing, etc while removing variability and cost from their process. No customizations. Scripted testing
for high risk (label printing); unscripted testing for medium and low risk features.
• COTS MES from a trusted, established vendor within the industry (ISO certified, robust SDLC, etc).
• Physical manufacturing process is not changing. Downstream controls to inspect/check product remain.
• Based on downstream controls and risk based assurance approach, device quality, device safety, and
patient/user safety risk have been mitigated. CDRH may accept filing in an annual report. Note – this is
developing policy evaluation.

Scenario 2: COTS MES Upgrade

• Intended use: Refer to scenario 1. No new features. Unscripted to verify no unintended issues (try to break system).
No customizations. The Physical manufacturing process is not changing.
• CDRH may accept filing in an annual report. Note – this is an ongoing evaluation this is developing policy
evaluation.

48
www.fda.gov
 Example: Annual Filing vs 30 Day PMA Notification
Scenario 3: COTS MES Upgrade
• Intended use: Refer to scenario 1. As part of upgrade, adding new features such as process timers (automating manual
function, e.g. timers for ovens). The Physical manufacturing process is not changing.
• No customizations. Scripted testing for high risk; unscripted testing for new medium & low risk features.
• Downstream controls to inspect/check product remain.
• Based on downstream controls and risk based assurance approach, device quality, device safety, and
patient/user safety risk have been mitigated. CDRH may accept filing in an annual report. Note – this is
developing policy evaluation

Why: FDA’s objective for these efforts is to enable safer and better medical devices for patients
and providers, by making it easier, faster, and more value-added for manufacturers to deploy
the necessary automation.

49
www.fda.gov
 Automated Computer System Validation Tools
Function Intended Use Examples
Software testing tool measuring system Used for testing the performance of new manufacturing *Loadrunner, Apache JMeter
behavior and performance under load automations under load
Automated functional graphical user Used for developing a test script based on user interactions to *Winrunner, Ranorex
interface (GUI) testing tool that allows a automate future testing of UI modifications
user to record and play back user interface
(UI) interactions as testscripts.
Bug tracking, issue tracking, and project Used for rapidly capturing issues and bugs found during assurance *Jira, Confluence
management systems. testing
Manage and track the application lifecycle Used for tracking and monitoring all stages of new IT system *Polarion ALM, PTC Integrity
development process. Includes, risk, test, implementations, throughout the lifecycle.
and the respective change control/approval
of applications
Dynamic web performance evaluation tool. Used for testing the performance of web-based User Interfaces *Dynatrace AJAX Edition, New
Relic APM
*All product trademarks, registered trademarks or service marks belong to their respective holders.

Manufacturer is using these tools to automate and supplement tracking and assurance testing for non-product
systems. These intended uses of these tools do not have a direct impact on device quality and device safety.

21
www.fda.gov
Qualification of Automated CSV Tools
Example 1 – Code Debugger
A CAPA automation system is being written in Java script and a developer tool is used to set up breakpoints and step through of
the code. Once the code is debugged all the debugger content is removed prior to the promotion of the code to the production
system. The debugger tool is not part of production or the quality systems.

Step 1: Identify where and how the Debugger will be used within your organization
Step 2: Determine if the intended use for automating part of production or the quality system
Consider the following in your decision, then capture the decision with rationale
• The off-the-shelf tool is not part of or integrated with the production or the quality system
Step 3: Assure the use within your organization
Assurance of the code debugger tool includes testing the tool for use within your organization.
1. Identify or create code to be debugged with known error types
2. Execute the debugger and verify that all expected error types manifested
3. Correct the errors in the code and execute the code to verify that the debugger produced executable code to meet
your use.
Step 4: Evidence
Record the intended use decision and rationale, as well as the acceptance conclusion for the tool.
Qualification of Automated CSV Tools
Example 2 – Automated Testing Tool
An ERP system has a load requirement and HP Loadrunner is used to simulate anticipated peak load of the production system.
The load testing results assures that the system can absorb the required user load. Then the automated testing tool used to test
load to a production system is not part of production or the quality systems.

Step 1: Identify where and how the testing tool will be used within your organization.
Step 2: Determine if the intended use for automating part of production or the quality system
Consider the following in your decision, then capture the decision with rationale:
• The testing tool is not the system of record of the product testing results
• The test tool does not alter the code within the production system
• The testing does not add any data to the production system
• The tool is not used for verification of Medical Device
Step 3: Assure the use within your organization
Assurance of the automated testing tool includes testing the tool for use within your organization.
1. Identify the type of testing results that will be achieved with the testing tool
2. Execute known test cases that represent a solid sampling of test types and conditions that will be encountered during use.
3. Ensure that the testing tool produced the testing results that were expected to meet the testing requirement of your organization
that will minimize defects being introduced into a production environment for your organization.
Step 4: Evidence
Record the intended use decision and rationale, as well as the acceptance conclusion for the tool.
Qualification of Automated CSV Tools
Example 3 – Defect management and ALM tools
A medical device company uses Polarion to automate the company’s CSV process including testing, defect management and other software life
cycle functionality in the support of implementation of a quality production system. The activities performed, and records maintained in Polarion
support the execution of the company CSV procedure and is part of the quality systems.

Step 1: Identify where and how a Defect Management and ALM tool will be used within your organization
Step 2: Determine if the intended use for automating part of production or the quality system
Consider the following in your decision, then capture the decision with rationale that validation is not applicable:
• ALM tool is used to execute company’s CSV process and does not alter the Production system data but automates part of quality system
• ALM tool is configured to automate company’s CSV process and does not impact nor interface with validated production quality systems.
• The testing does not add any data to the production system
Step 3: Assure the use within your organization
FDA considers that impact to the quality system does not present a direct patient or user safety risk. Assurance of the automated testing tool
includes testing the tool for use within your organization.
1. Identify specific functionality or process that the ALM or Defect management tool will automate by creating specific functional level requirements.
2. Execute known test cases that represent functional requirements under a variety of conditions that represent organizational use.
3. Ensure that the testing results produced the desired outcome to a level that provides full confidence in the tool(s) functionality to meet the
intended use of an ALM or Defect management tool within your organization.
Step 4: Evidence
Record the intended use decision and rationale, testing results, as well as the acceptance conclusion for the tool
Complaint Handling Spreadsheet example

Intended Use: Extract complaint data from plant local systems & identify complaint trends across regions in injectables sources
Example Requirement: Global QM shall be able to open read-only complaint data extracts from “xyz” secure storage location.
Risk Level: LOW (patient), a configured implementation method
Spreadsheet Tested: Spreadsheet X, Version 1.2
Test Type: Unscripted Testing
Goals:
• Ensure Global QMs are able to open complaint data extracts from “xyz’ secure storage location (data retrievability)
• Ensure complaint data extracts are read-only (data protection testing) and “xyz” storage location is accessible only to
authorized individuals (data security testing
Assurance activities:
• Performed data retrievability, data security and data protection testing as outlined in the testing goals. Included positive as
well as negative test conditions for testing if unauthorized individuals can access the data.
Conclusion: No errors observed.

The intended use was identified to be low patient risk and rapid exploratory testing of specific functions were performed. The
resulting record quickly identifies the intended use, what was tested, how it was tested, the test objective, who performed the
testing, and conclusion on validation activities.
Risk Based CSV Example:
Learning Management System (LMS)

A medical device firm applies Risk Based Validation to an off the shelf LMS. Qualifying the vendor
then applying risk to the feature level allows for much less documented verification activities.

Ad Hoc
Basic Assurance / Ex: Usability Features – training notifications,
Testing
Low Risk Features overdue training lists, curricula assignments.
80%

Unscripted
Ex: Capture evidence of training completion by
Medium Risk Features Testing
entering username & password.
20%

Scripted
High Risk Features No High Risk Features Testing
0%
Risk Based CSV Example:
Non-Conformance & CAPA Process

A medical device firm applies Risk Based Validation to an off the shelf CAPA System. Qualifying the vendor
then applying risk to the feature level allows for much less documented verification activities.

Ex: Usability Features – required data entry from Ad Hoc

Basic Assurance /
optional data entry, attachments of objects, Testing
Low Risk Features
system workflow, non conformance initiation. 30%

Unscripted
Ex: Electronic Signature Features – audit trail,
Medium Risk Features Testing
meaning of signature (review, approval).
50%

Ex: Product Containment – NC is initiated for

Scripted
product outside of the company’s control, then the
High Risk Features Testing
system prompts the user to identify if a product
20%
recall is then needed.