Presentation title Date 1

Understanding the changes to

the risk standard for medical
devices: EN/ISO 14971:2012
Peter Bge
Informa Post Market Surveillance and Vigilance conference
Amsterdam, 24. -26. February 2014
 Presentation title Date 2

Background

Peter Bge
>20 years working with development processes within medical devices
>15 years working in ISO and IEC with
ISO14971 Risk Management,
IEC6366 Usability
IEC60601-1 Safety of electrical medical devices
 Presentation title Date 3

A gruesome story,
with a happy end, starring the PMS-hero
1. Why the risk standard has been amended
2. Examining the content deviations in Annex
ZA, ZB and ZC and how this will impact
medical device and IVD manufacturers
3. Clarifying how medical device manufacturers
should adapt to meet the requirements of
the standard
4. Some open questions
 Presentation title Date 4

1. Why the standard has been amended: History

In the MDD from 1993 Essential Requirements were

phrased ambiguously regarding risk mgt.
However consensus until the EU Commission in 2011
raised a warning to deharmonise ISO14971.
CEN communicated with the EU Commission, initially
aiming at convincing the Commission,
Having failed in that, CEN wrote Annex Z explaining,
where the Commissions saw it not meeting the Essential
requirements of the MDD, spelled out as 7 Content
deviations.
Resulting in the European EN14971:2012, with Annex
Z being the only difference to ISO14971:2007.
The ISO JWG 1 for ISO14971 has continually been
following the talks with the Commission and given
scientific advice to CEN. But has no formal role.
 Presentation title Date 5

1. Why the standard has been amended:

The resulting EN14971
EN ISO14971:2012 is one of a kind:
An Annex is not normative - but this Annex Z
contains normative requirements,
The content deviations are stated to be from the
original 1993 MDD - but for most readers, they
are new requirements.
Standards usually become mandatory after e.g. 3
years - this became mandatory 19 years ago (!)
Lets hope it will remain a one of a kind.
Some of the deviations are raising the bar/
adding to requirements,
some seem to conflict with the risk mgt principles
of ISO14971, the industry and risk mgt. science.
 Presentation title Date 6

The question of the

Essential Requirements 1-4
The existing ER do not seem to
express the requirements in Annex Z well.
The proposed new ones have ben fitted to do that
So now its a bit more clear
But are they the right ones?
 Presentation title Date 7

1. WhyDifferences between US and EU

FDA (U.S.A.) and EU

New requirements are also coming from FDA
Unlike EU they target using internationally recognized
standards more, not less
Unlike EU FDA is actively involved in standardization
E.g. ISO14971 & IEC62366.
Unlike EU FDA seems more committed to discussing
practical issues in industry regarding new
requirements.
Part of the standardization process
FDA seems more interested in raising the bar for data
(related to e.g. risk mgt.) than principles.
Principles are part of the standardization process
 Presentation title Date 8

2. The impact of content deviations in Annex Zx

Content deviation Impact. Issues?

1.
Treatment of negligible risks: A statement in - i.e. also the
manufacturer must take all risks the Risk Mgt acceptable/small ones.
into account when assessing Report may be How to demonstrate that
Sections 1 and 2 of Annex I of sufficient you have identified all risks?
the directive.
2.
Discretionary power of Update of RM How to demonstrate
manufacturers as to the Report may be that you have reduced
acceptability of risks: all risks sufficient all, potentially many
have to be reduced as far as small, risks?
possible and that
all risks combined, regardless What does combined
of any "acceptability" mean?: Add? - Fault
assessment, need to be balanced, Tree?
together with all other risks,
against the benefit of the device. Is this the right focus?
 Presentation title Date 9

The question of harmonized standards

Do the content deviations also apply to risks that

have been addressed by meeting a standard?
E.g. IEC60601-1 series or ISO10993 series.
If yes, what then is the point in having harmonized standards?
If not, is the concept of risk in MDD different from that in
ISO14971? Suggestion:
14971- Risk can be reduced to an acceptable level, but not to zero..
MDD Risk cannot be reduced to acceptable level, but may be
eliminated by meeting a harmonized standard.
 Presentation title Date 10

2. The impact of content deviations in Annex Zx

Content deviation Impact. Issues?

3. 1. If disregarding that As Far As

Risk reduction "as far as
possible" versus "as low as
reasonably practicable":
manufacturers
and Notified Bodies [!] may not
apply the ALARP concept with
regard to economic considerations.
Update of Possible usually is reserved
RM Report nuclear plants-safety (i.e. issue
may be with the science of RM),
sufficient 2. And disregarding the economic
considerations of society (who is
the ultimate payer), then:
3. You can still apply ALARP on
technical considerations (State of
the Art).
 Presentation title Date 11

2. The impact of content deviations in Annex Zx

Content deviation Impact. Issues?

4.
Discretion as to whether a risk-benefit Update of RM If the Clinical Evaluation
analysis needs to take place: the Report may report is used.
manufacturer must undertake the risk- be sufficient Makes sense for injection
benefit analysis for the individual risk devices.
and the overall risk-benefit analysis Makes less sense for devices
(weighing all risks combined against the used in complex medical
benefit) in all cases. procedures: Then it is truly a
new requirement.
2. The impact of content deviations in Annex Zx
ISO 14971, 6.2: Content deviation Impact. Issues?
The manufacturer
shall use one or
more of the 6. Some think its a new
following risk Deviation as to the first risk Update of requirement.
control options in control option: RM Report How to demonstrate
the priority order Eliminate or reduce risks as far may be that youve done all?
listed: as possible (inherently safe sufficient
design and construction)"
a) inherent safety 5. Some think its a new
by design; Discretion as to the risk control Update of requirement.
options / measures: the RM Report How to demonstrate
b) protective manufacturer must apply all the may be that youve done all?
measures in the "control options" and may not sufficient
medical device stop his endeavours if the first
itself or in the or the second control option has
manufacturing reduced the risk to an
process; "acceptable level" (unless the
additional control option(s)
c) information for do(es) not improve the safety).
safety.
 Presentation title Date

The question of
13

as far as possible vs. ALARP?

At some point tests will not demonstrate any effect of

additional risk control measures
Shall the manufacturer still do all possible that MAY reduce risk?
Difficult to see happening in the real world
Or only all possible that demonstrably reduces risk?
I.e. if data from e.g. test demonstrates a lower probability or severity of harm or both.
State of the art, e.g. based on market data and literature, is till needed
But if you have to do as much as possible
For all risks, big or small,
Until state of the art makes it impossible to do more,
Then its not ALARP that has been abolished
Its all ALARP!
2. The impact of content deviations in Annex Zx
Content deviation Impact. Issues?

Information of the users influencing If warnings have 1. Difference

the residual risk: reduced risk between
b) However, the last indent of scoring (e.g. RPN), warning and risk
Section 2 of Annex I to Directive then analysis has to disclosure?
93/42/EEC says that users shall be be redone.
informed about the residual 2. Some warnings
risks. This indicates that, according To justify safety of are required by
to annex I to 93/42/EEC and existing products, standards
contrary to the concept of the one way is PMS
standard, the information given to data analysis. 3. Existence of
the users does not reduce the Probably the most Instructions
(residual) risk any further. effective and easy For Use
c) Accordingly, manufacturers shall way. (prescriptive
not attribute any additional risk description of
reduction to the information tasks) will still
given to the users. reduce risk!!!
 Presentation title Date 15

3. How manufacturers should adapt

Go through the Content Deviations

- Identify gaps in your documentation.
Some Content Deviations may be fixed with
adding statements to the Risk Mgt. Report.
Some deviations can only be addressed by
statements, not by data (e.g. #6, reducing risk as far
as possible)
You may need to introduce risk benefit analysis.
#7: Most will have to remove warnings from risk
scoring (RPN):
Potentially biggest issue.
 Presentation title Date 16

3. How manufacturers should adapt

Content deviation 7 usually requires more: Usually

update of Risk Analysis and worse - labelling:
1. Remove warnings as a RCM from the risk scoring (RPN
calculations).
2. Do not remove them from the actual IFU/labelling.
1. May be required by standards or even legislation

2. And maybe they actually do have a risk reducing effect!
3. Re-calculate the risk score (e.g. RPN).
1. Identify any seemingly INTOLERABLE risks.
2. Assuming that PMS process has not indicated safety issues,

3.
are they INTOLERABLE? No.
You may therefor do a new risk ESTIMATION the rationale
=
being that the risk estimates have not been linked as close to
PMS data, as they should have been
 Presentation title Date 17

A review of sources for risk estimates

ISO 14971, 5. NOTE 6
Information or data for estimating risks can be
obtained, for example, from:
a) published standards;
b) scientific technical data;
c) field data from similar medical devices already in use
including published reported incidents;
d) usability tests employing typical users;
All may be used, but
e) clinical evidence; some are more
f) results of appropriate investigations; practical or convincing
g) expert opinion; and than others
h) external quality assessment schemes.
If available,
D 3.2.1. Seven approaches are commonly market feed back
employed to estimate probabilities:
use of relevant historical data;
data is fastest,
prediction of probabilities using analytical or simulation cheapest and most
techniques; convincing
use of experimental data;
reliability estimates; Agree your approach
production data; with your Notified Body
post-production information; or
use of expert judgment.
before
 Presentation title Date 18

Review risk analysis on marketed products

1. A ISO14971 compliant system will have established

links between
Risk, incl. Failure Mode: How a failure is observed
(should be searchable in e.g. a FMECA, but may not be).
- and codes for Complaints
leading again to found Technical Errors
2. Risk Analysis: Do ERRORS happen?
PMS: Complaints indicate presence of ERRORS
3. Risk Analysis: Do ERRORS lead to harm?
With what SEVERITY?
PMS: Adverse-events indicate medical consequences of
ERRORS
PMS: Complaints without Adverse Events may indicate
low severity scenarios
4. In less-than-perfect complaint handling this may
require expert judgment
(i.e. meetings between Risk and PMS people)
 Presentation title

Handling the critical risks

You may end up with 2 -10 risks, Date 19

with both complaints and adverse P5

events.
Users will face the ERROR from P4
time to time.
And some will experience HARM P3 X
You have monitored these risks,
and determined that they are P2 X
acceptable.
The remaining risks: P1 X X
Should as per this analysis, be
lower. S1 S2 S3 S4 S5
Update your risk analysis
accordingly
You may have done this to
perfection (as you should acc. to
ISO14971), or you may not
 Presentation title Date 20

Critical risks in Information For Safety

For residual risks that are judged acceptable, the
manufacturer shall decide which residual risks to
disclose
and what information is necessary to include in the
accompanying documents (ISO14971, 6.4)
I.e. the critical risks
But may not describe all needed elements:
1. Error: What not to do (warning),
2. Resulting probability + +
3. Consequence (harm) and
4. Severity.
E.g. Do not press button, as this may lead to burns,
requiring healthcare intervention.
Impact
This old requirement is NOT part of Annex Z
But weaknesses if any, become clear
And may if applicable be the most costly part!
But thats another story
 Presentation title Date 21

Thanks for listening!