Risk Management 101
Risk Management 101
orielstat.com 800.472.6477
From a distance, risk management seems straightforward. You have a device,
evaluate its potential risks, mitigate those risks, monitor them over time, and you’re
done. Seems easy, right? Ah, if only life were so straightforward. The reality is that
risk management is one of the more complex aspects of regulatory compliance,
simply because risk comes in so many flavors and perceptions of severity, and
probability can be interpreted quite differently.
The thing that makes risk management tricky is that we often don’t have enough
real-world data to accurately quantify risks, especially for new devices. Fortunately,
there is a systematic process you can establish to estimate, evaluate, control, and
monitor risks. Before we get into that, let’s step back and talk about the regulations
and standards that dictate how you should approach risk management.
ESTIMATE
RISKS
MONITOR
EVALUATE
THE
RISKS
CONTROLS
CONTROL
RISKS
2 orielstat.com | 800.472.6477
Why is risk management needed? Simply put, we have a collective interest in ensuring that medical
devices are safe and effective. Risk management is not optional – it is a regulatory requirement world-
wide. The US FDA mandates it in the Quality System Regulation (21 CFR Part 820). Europe requires
it in the new Medical Device Regulation (MDR 2017/745). Likewise, Japan, Canada, Australia, Brazil,
and all other major markets require the application of risk management, which is either referenced in
their national regulations and/or ISO 13485:2016.
• ISO 14971:2007 – The US FDA and most other markets recommend this version of the standard to
meet national risk management requirements.
• ISO 14971:2012 – This version is required to meet CE Marking requirements for medical devices
sold in Europe. It differs only in the front matter describing how ISO 14971:2007 deviates from the
device directives in Europe.
If you are just getting started implementing risk management for your company, purchase the
ISO 14971:2012 standard and its guidance ISO 24971:2013. You will also want to buy and read the
ISO/TR 24971:2013 standard. It is brief but provides excellent guidance for dealing with specific
areas of ISO 14971. Both are copyrighted documents and you can purchase them online from ISO.
2000 2007
1998 2012
2001 2009
800.472.6477 | orielstat.com 3
Sections of ISO 14971
Although risk management can be complex, the main body of the ISO 14971 standard is a scant
14 pages and consists of 9 clauses:
1. Scope
2. Terms and conditions
3. General requirements for risk management
4. Risk analysis
5. Risk evaluation
6. Risk control
7. Evaluation of overall risk acceptability
8. Risk management report
9. Production and post-production information
4 orielstat.com | 800.472.6477
Basic steps in the medical device risk
management process
So, where to begin? It helps to think about risk management as a process that starts with a plan.
While the end deliverable is a report, your work in controlling risk is never done. We will talk in detail
about each of these areas later, but here are the steps.
RISK ACTIVITIES
Risk Analysis Risk Management Risk Control
and Evaluation Tools • Risk Reduction Options
Risk Risk
Management
Plan
> • Intended Use
• Risk
Identification
• FTA
• FMECA
• Others
• Implementation
• Residual Risk > Management
Report
• Risk-Benefit Analysis
• Estimation • Verification and Validation
• Evaluation
• Create a risk management plan: Document activities that will take place, assign responsibilities,
determine risk review requirements, establish risk acceptability levels, plan verification activities,
and plan production/post-production activities.
• Assemble your risk management team: Assemble a qualified team of people who know how
your device is constructed, its manufacturing processes, how it is used in the field, etc.
• Use risk analysis tools to identify risks: Choose the tools you will use to measure risk (discussed
more later) and then use them to identity risks posed by your processes, users, suppliers, mainte-
nance tasks, shipping, production equipment, etc.
• Weigh the risks versus the benefits: This is fairly self-explanatory, but the end goal is to ensure
that the medical benefits of your device outweigh residual risks.
• Eliminate or mitigate risks: The goal here is to reduce risks to an acceptable level. We’ll talk more
about risk reduction later and address how this varies between the 2007 and 2012 versions of
ISO 14971.
800.472.6477 | orielstat.com 5
Creating your risk management plan
It’s important to understand that risk management is a process and a project. In other words, risk
management is much more than a periodic analysis of product risks. Clause 6.1 of ISO 14971 states
that you must have an ongoing process in place to analyze, evaluate, and control risk. This plan
outlines the process of how you will conduct risk management, and it becomes part of your risk
management file. Risk management planning may have two levels.
There are several activities that should be part of your risk management plan, and we will talk more
about them later. First, you need to define the scope of what you will be evaluating, including a
detailed device description and its life cycle. This is also the time to clearly lay out your risk accept-
ability criteria, verification activities, and production and post-production plans. Your plan should
also clearly define the responsibilities of your team.
6 orielstat.com | 800.472.6477
RISK ANALYSIS RISK ESTIMATION RISK ASSESSMENT RISK EVALUATION
Process of
comparing
Systematic use of Process used to
Overall process the estimated risk
available assign values to
comprising a risk against given risk
information to
identify hazards
and estimate a risk
+ the probability of
occurrence and
severity of harm
+ analysis and risk
evaluation
= criteria to
determine the
acceptability of
(ISO 14971 2.18)
(ISO 14971 2.17) (ISO 14971 2.20) the risk
(ISO 14971 2.21)
Put yourself in the shoes of the user or patient. What could go wrong during typical use situations?
Could the device be misused in a way that would cause harm? Are there other similar products on the
market? What has gone wrong with them? FDA databases, published journal articles, online product
reviews (consumer devices), and user interviews are good sources for such information. Your existing
design control program can also yield useful information and Annex D is a good starting point. The
extent to which you perform this analysis largely depends on the risk classification of your device.
ESTIMATE
IDENTIFY KNOWN ESTIMATE ESTIMATE
PROBABILITY
OR FORESEEABLE PROBABILITY SEVERITY
OF HAZARDOUS
HAZARDS OF HARM OF HARM
SITUATIONS
800.472.6477 | orielstat.com 7
Hazards and harms
After you have identified hazards and associated harms, you need to estimate the probability that
they will occur. Not all hazards will result in harms.
Annex E of ISO 14971 shows the relationship between hazards, hazardous situations, harm, and foreseeable
sequences of events.
You need to perform an in-depth analysis, looking at the role of the user and use environment. Here’s
an example. Let’s suppose you make a blood glucose meter. Your product displays the most important
readings in very large text. If you examine the screen while sitting in your office, you might assume
that the probability of a misread by the user is quite low. But what happens when you take it outside
into bright sunlight? Is the display screen highly reflective? Can you clearly read everything using
sunglasses? How about in low light? Is the battery meter clearly visible and does it provide adequate
warning of battery depletion? These are potentially hazardous situations and your mission is to
estimate the probability of those situations. Regulators expect you to anticipate these issues. Never
blame the user.
Let’s go back to the battery indictor issue we just mentioned. In this case, the “hazard” might be a
battery indicator that is too small on the LED screen and without any supplemental warning light to
signal that the battery is very low. A “hazardous situation” that might result involves users who need
to check their insulin level right away, only to find their glucose meter is out of power. If the users do
not have a way to recharge the meter for 2 hours, they may simply guess how much insulin they need
based on how they are feeling. The harm that could result is hypo- or hyperglycemia caused by
improper dosing of insulin. This risk can be mitigated by making the battery meter larger, and/or
by adding a supplemental visual or audible indicator by which the battery warns users that
recharging is needed.
HAZARD
+ FORESEEABLE
EVENTS + HAZARDOUS
SITUATIONS = ESTIMATE OF
RISK
8 orielstat.com | 800.472.6477
Estimating the probability of harm
Risk is a combination of the severity of a harm and the probability that it will occur. Clause 4.4 of
ISO 14971 requires you to estimate the probability of harm. But how is that done? You can reference
historical data or published FDA data, and try to better understand typical use scenarios. A qualita-
tive probability table similar to the one shown below will help you tackle this process of evaluating
potential hazardous situations. You can also do something similar using a numerical system, rating
the severity of the harm. Annex D provides some guidance on risk analysis concepts, including risk
estimation, but you can create your own scale and descriptors as long as you define them in your risk
management procedure. Keep in mind that harms can have levels of severity. An example of this
would be burns.
Frequent
Happens with
almost every use CONTROL UNACCEPTABLE UNACCEPTABLE UNACCEPTABLE UNACCEPTABLE
of the device
Probable
PROBABILITY OF OCCURENCE
Occurs the
majority of times CONTROL CONTROL UNACCEPTABLE UNACCEPTABLE UNACCEPTABLE
but not with every
use
Occasional
Occurs with
increased ACCEPTABLE CONTROL CONTROL UNACCEPTABLE UNACCEPTABLE
frequency
Remote
More than one
occurrence per ACCEPTABLE ACCEPTABLE CONTROL UNACCEPTABLE UNACCEPTABLE
year but still
unlikely
Improbable
Less than one
occurrence per ACCEPTABLE ACCEPTABLE ACCEPTABLE CONTROL CONTROL
year; isolated
events
800.472.6477 | orielstat.com 9
Ultimately, risk estimation should be viewed as a data-driven process. Gather as much quantitative
information as possible from your complaint-handling files, published standards, technical data,
clinical data, results of investigations, expert opinion, field data, MDRs, and test data. You can docu-
ment it using Excel, software tools, or a simple list.
Using this list as a starting point, research the various options and find the best fit for your situation.
Typically, it is optimal to use more than one tool and a combination of top-down and bottom-up tools.
That last question about the acceptability or residual risk is important. There is a significant difference
between ISO 14971:2007 and ISO 14971:2012 about this issue, more specifically ALARP (as low as
reasonably possible) and AFAP (as far as possible).
10 orielstat.com | 800.472.6477
Commonly Used Risk Evaluation Tools
BOTTOM-UP TOOLS HOW IT IS USED
Preliminary Hazard Analysis “What if” analysis that takes a hazard and traces it to harm. Useful early
(PHA) in the risk management process.
Failure Mode and Effects “What if” analysis that takes a failure and traces it to an injury or hazard.
Analysis (FMEA) Often used in design and process phases. Focuses on one piece of the
puzzle, but don’t rely on it alone for your risk management process. Best
for manufacturing and use instructions.
Failure Mode, Effects, and Builds on the FMEA model but adds risk evaluation, including severities
Criticality Analysis (FMECA) and probabilities.
Fault Tree Analysis (FTA) Starts with failure and works back to the component. Focuses on the big
picture, unlike FMEA. Good choice for design activities. Use with FMEA
for new design technology unless failure modes are unknown.
Turtle Method Starts with a process and examines factors that influence the process
such as training, equipment, procedures, installation, etc.
Hazard Analysis and Critical Identifies various errors and hazards in production processes that can
Control Points (HAACP) cause finished products to be unsafe. Designs measurements to reduce
risks to a safe level.
Hazard and Operability Analysis Assumes accidents are caused by deviations from design or operating
(HAZOP) intentions. Uses keywords to focus attention on specific aspects of
design intent or associated process condition.
800.472.6477 | orielstat.com 11
ALARP or AFAP?
An important part of the risk analysis process is to ensure that you do not introduce
new hazards in your quest to eliminate or minimize hazards. FDA describes their
expectations about risk-based decisions. In 21 CFR Part 820, FDA states that if any
risk is judged to be unacceptable, it should be reduced to acceptable levels by the
appropriate means, which may include a redesign or warnings. Similarly, ISO
14971:2007 espouses the general concept of reducing risk to ALARP (as low as
reasonably possible). The key word here is reasonable, which is open to interpreta-
tion. In practice, within reason includes a consideration of “state of the art” from a
technical perspective. Although Annex D does mention the “cost of further reduc-
tion” in the definition of practicability, it is never acceptable to trade off device safety
against cost, so don’t even put that in your justification. The rationale for these
decisions must be documented in your risk management file.
ISO 14971:2012 – the version of the standard required for medical devices sold in
Europe – embraces the concept of reducing risk AFAP (as far as possible), which
leaves no room for interpretation. It ignores economic arguments altogether. If a
risk exists, you are required to reduce it to acceptable levels and explain why it
cannot be reduced further.
12 orielstat.com | 800.472.6477
Risk Management Flow Chart (ISO 14971:2012)
RISK PLAN
ANALYSIS
START Risk Management Intended Use: Identify Identify Known or Foreseeable Hazards (4.3)
AND
NO
RISK
Is risk reduction
necessary? (5)
YES
Is the risk NO
reducible? (6.2)
YES
NO
NO Are all
identified hazards
considered? (6.7)
YES
RESIDUAL RISK
EVALUATION
OVERALL
Do medical
NO Is overall residual benefits outweigh NO
risk acceptable? the overall residual
(7) risk? (6.5)
YES
REPORT
UNACCEPTABLE
Review production and post-production information (9)
INFORMATION
PRODUCTION
Is
POST-
YES reassessment of NO
risk necessary?
(9)
800.472.6477 | orielstat.com 13
Hazard does not equal harm
Risk management is the process of reducing risk. However, sometimes the likelihood of harm result-
ing from a hazard is quite low and mitigating that hazard may not provide any tangible reduction in
risk – in fact, it may diminish your device’s benefits. Here’s a quick example. Let’s say you produce an
IV bag stand. To reduce the likelihood of it getting knocked over, you could make the caster base legs
25% wider. However, doing so would introduce a new hazard: wider base legs increase stability but
also increase the tripping hazard. If the likelihood of the stand getting knocked over is low, you may
actually increase overall risk. To avoid making the wrong decision, evaluate risks using a disciplined,
planned process. In your risk management file, document the hazards you identify and the rationale
behind the decision to control or not control those risks. Remember, you cannot trade off safety
for cost!
Your risk management file must include evidence that you have conducted the following for each
identified hazard:
• Risk analysis
• Risk evaluation (probability, severity)
• Implementation and verification of risk control measures
• Assessment of the acceptability of residual risk
Annex I of the European Medical Device Regulation (2017/745) says that you should “reduce risks as
far as possible” without adversely impacting the benefit-risk ratio. This is in line with the concept of
AFAP introduced in ISO 14971:2012 and discussed earlier. The US FDA also publishes an excellent
guidance document discussing the risk-benefit evaluation process for medical devices. Ultimately,
FDA recommends that you take the following factors into account:
• Type of benefit – quality of life, relief from symptoms, reduced probability of death, etc.
• Magnitude of benefit – anticipated change in condition or clinical management
• Probability of benefit – can be based on prior investigations, demographics, health status, etc.
• Duration of benefit – curative or repeated interventions required
• Availability of alternatives – safety and effectiveness of other options
14 orielstat.com | 800.472.6477
To ensure that you do not go overboard in analyzing residual risks, it is important to establish a
systematic process and focus on the risks that are within your control. Therefore, choosing the right
risk management team is extremely important. You need to assemble people who fully understand
how your device is manufactured, distributed, and used – someone without any knowledge of how
your device is manufactured will not be able to foresee scenarios that could create hazardous situa-
tions. If you are missing expertise in one area, you may miss potential harms to evaluate and control.
DEVIATIONS
RISK MEDICAL
ASSESSMENTS BENEFITS
Risk
Management
Report RISK-BENEFIT
CONCLUSIONS
SUMMARY
RISK OF
CONCERN
800.472.6477 | orielstat.com 15
Post-production risk management
Once your device is for sale on the market, congratulations: your risk management work is complete.
Ok, just kidding – your work is never complete! Risk management is a continuous process. Typically,
you will find yourself dealing with two types of post-production issues:
Once your device is for sale on the market, congratulations: your risk management work is complete.
POST-PRODUCTION ISSUES
Ok, just kidding – your work is never complete! Risk management is a continuous process. Typically,
INCIDENT-DRIVEN
you will find yourself REVIEW-DRIVEN
dealing with two types of post-production issues:
Requires your immediate attention Discovered during data analysis
Your risk management process should document both pathways for analysis. As part of your ongoing
efforts, you should be evaluating complaints, incidents, product failures, and design process changes
for potential safety impact. You will also take into account any changes in installation, use, and servic-
ing. Are previously unrecognized hazards present? Is the estimated risk no longer acceptable? Is the
original assessment still valid? Possible triggers include:
• Design/materials changes
• Manufacturing changes
• Vendor changes
• Complaints
• Medical device record
• Incidents
• Near-incidents
• Malfunctions
You are required to analyze all incidents, near-incidents, and malfunctions to categorize their risk
level. Your assessment must be documented, becomes part of your risk management file, and may
result in a corrective and preventive action (CAPA) or having to file a vigilance report or other regula-
tory notification.
16 orielstat.com | 800.472.6477
Want to learn more?
Call 1.800.472.6477 or visit www.orielstat.com
Oriel STAT A MATRIX has been assisting medical device manufacturers with compliance issues for
over 50 years. If you are just starting out to set up a risk management system in your company, you’ll
definitely want to check out our intensive training course on ISO 14971 and risk management. Our
expert consultants can also provide hands-on risk management support.
800.472.6477 | orielstat.com 17
MT-007-0918