Tuesday, December 18, 2012

EN ISO 14971:2012...seriously?

The revision 2012 of the harmonized standard EN ISO 14971 pertaining to risk management for medical devices has been implemented within the European regulatory framework on August 30, 2012.
Although the core of the text of the standard itself is unchanged from the ISO 14971:2007, the update of the Annex ZA has introduced major confusions and headache for manufacturers and certainly for Notified Bodies. In particular, this revision bans the use of the 'As Low As Reasonably Practicable' (ALARP) approach from the risk acceptance process and basically considers that the labelling is not a risk control measure that may decrease the occurrence of a risk. This is what I call A change!

The changes brought by the revision of the EN ISO 14971:2012 harmonized standard could have easily gone unnoticed. The standard indeed fully endorses the text of the current ISO 14971:2007 and the updates only pertain to the informative Annex ZA. While the Annexes ZA has very limited normative impact (for information only), their regulatory impact is however crucial. As always, the devil is in the details... 

The Annex Z ensures the regulatory binding between the Essential Requirements of the Medical Devices Directive 93/42/EEC (MDD) and the use of harmonized standard to support compliance, in direct alignment with the Article 5 of the MDD. Said differently, the Annex Z of an harmonized standard corresponds to the current thinking of the European Commission about the relevance of a standard to respond to the Essential Requirements of the MDD. Usually, the interpretation of the European Commission is direct and no specific limits to the standard are made. The Annex ZA (the version 2012 includes 3 Annexes ZA, ZB and ZC - one for each Directive) introduces a particular degree of interpretation on how a manufacturer should apply the standard to respond to the Essential Requirements of the MDD. The direct impact is that a manufacturer thoroughly claiming the compliance to the ISO 14971:2007 without including the requirements of the Annex ZA may not be compliant with the MDD!

While some Notified Bodies have issued gap analyses to help (alert?) manufacturers, the application of the revised standard remains very obscure. In particular, the LRQA (No. 0088) has published an interesting document. Among the changes described by the Annex ZA, the following are of particular interest: the removal of the ALARP approach and the loss of impact of the labelling to mitigate a risk.

The end of the ALARP approach

The ALARP approach is no longer accepted and neither Notified Bodies nor manufacturers should consider a risk as acceptable with regards to economic considerations. Risk reduction "as far as possible" applies instead. Additionally, all risks - regardless of their initial risk level - need to be individually reduced as far as possible and balanced against the expected benefits.
Consequently, the risk reduction should be performed regardless of the cost of the risk reduction measures. But where is it supposed to end and what could limit a Notified Body from requesting a manufacturer to further mitigate a risk with control measure costing millions?  

In practice, an approach of risk reduction strategy performed together with a systematic assessment of the benefits/risks ratio should apply. It will then be a matter of interpretation and of application of a pragmatic approach. 

While the valuable intent of the European Commission was certainly to force manufacturers to preserve patients safety at any price, the interpretation of the Annex ZA will probably end into an 'As Far As Possible but Nevertheless Pragmatic' approach: the AFAPNP is born!

The labelling: do as if nobody reads it

According to the new Annex ZA, the occurence of a risk cannot be reduced by the labelling. The impact is typically that manufacturers cannot reduce the quotation of the risk only mitigated by a warning in the IFU... Putting this requirement in the perspective of (i) the removal of the ALARP approach and (ii) the 2nd Essential Requirement of the MDD which requests the mitigation by design, then protection and finally labelling, this means that a risk only mitigated by the labelling could never be accepted, unless a manufacturer drastically changes the design of his device regarldess of the cost. Let's illustrate this with the risk of allergic reaction to a material of an implant...People have allergies, so how can a manufacturer prevent from the risk of anaphylaxis (maximum severity) if it is not by the labelling? Should a surgeon instinctively know the composition of a device? Should a manufacturer change the materials of his device at any cost? In this case, should we immediately remove biological products from the market considering the potential allergies of the population? In practice, manufacturers will mitigate the risk by conducting hypersensitivity testing or the use of materials known for their low allergic potential. But still, people have allergies and only the IFU can provide the relevant information for the safe use.

Again, I fully appreciate the valuable intent of the European Commission to force manufacturers not using the IFU as the ultimate tool to protect themselves and apply the strategy 'it is in the labelling so I am covered'. But the downside of this restriction is that it reveals an admission of failure: healthcare professionals do not read the IFU and do not pay sufficient attention to the labelling. While I personally do not read the booklet supplied with my TV, I would expect licensed healthcare professionals to collect sufficient knowledge on the electromedical device they are using or on the material they are implanting, especially upon consideration of the evolutions and variety of this sector.

My personal opinion is that the IFU is important and supports both instructive and legal values. Therefore, a pragmatic approach should again be observed when it comes to the mitigation by the labelling and the systematic assessment of the benefits/risks ratio for each individual risk should apply.

7 comments:

  1. The labelling issue specifically excludes a risk reduction assignation via labelling but not that you cannot assign labelling as a control option - you just cant assign a reduction in the risk score as a consequence - in the example provided - anaphylaxis due to biological material allergy - the risk response would be labelling (this is still appropriate) but the RPN or risk rating could not be documented as reduced. Justification as to why labelling of this risk is considered 'AFAPNP' should be provided.

    The interpretation of the labelling risk control measure is controversial to say the least - long established principles of risk management denote a 'risk mitigation action' as an action (this is a key word here) taken to reduce severity or likelihood of occurrence. Labelling and warnings are considered a reduction of likelihood and would be acceptable risk mitigation actions. So the interpretation by the Commission not only provides an additional burden to manufacturers but is technically (by the principles of risk management)incorrect.

    The revised standard is however, here to stay for the forseeable future, so its impact will be heavily felt.

    ReplyDelete
  2. Thank you Peter, good comment. I agree with you: one can like it or not, the application of this harmonized standard is mandatory now that it has been published in the Official Journal of the EU. Considering that the update of the standard was almost left unnoticed, let's bet that manufacturers will painfully measure the impact of this update for their future Technical Files / Design Dossiers submissions. Hopefully Notified Bodies will accept a relative transition period. We should also hope that the requirements for mature products already on the market will not be too high - otherwise it will really be a documentation exercise for the sake of documentation.

    ReplyDelete
  3. Chiming in this late because you should have more of us agreeing with you. One of my clients submitted a Design Dossier update for shelf-life extension and minor IFU changes in October 2012, and was tortured by their NB into declaring conformity to the revised standard. It was indeed documentation for the sake of documentation. They did accept a revised Risk Management Report with language asserting that cost was not used as a determining factor in risk reduction, and that labeling was not used to reduce assumed occurrence rates.

    Since then, I've indulged in careful wording to avoid the "labeling" trap. For example, if appropriate device sizes are required to reduce risk, the response is "device is provided in appropriate range of size options based on...".

    ReplyDelete
  4. AnonymousJuly 08, 2013

    This revision is not a good one. medical device products are not going to improve as a result of this change. there will be more regulatory work, but not better designs since better designs were never stimulated by these kinds of requirements.

    This standard now misses the point. risk assessment and management is important and essential, but there is now motivation to avoid it, bypass it, and do it only as a paperwork exercise in a separate room from the engineers and testers.

    ReplyDelete
  5. A comment on the risk reduction by labelling. The annex ZA corresponds to the 93/42. But let's look at the annex ZC corresponding to 98/79 for in vitro devices. The essential requirements on the risk reduction by labelling are strictly identical in the two directives (Annex 1, part A, section 2, last bullet point), but the part concerning in vitro devices was not appended on the 14971:2012. So I can still reduce risk by using labelling on my in vitro medical devices, but cannot on my medical devices...

    I may understand this discrepancy if we consider that a labelling is intended for a patient, but absolutely not if the labelling is intended for the medical professional using a device... And anyway, I would rather add to the norm some requirements on the visibility of labellings and training of the users than that...

    ReplyDelete
  6. Hmmm, wee were audited Jan 2013 by a major NB, without any changes to out RMF or a mention of the 14971:12 Z annexes. I guess they were busy with the massive Ed3 retro-compliance. But, thanks for the info, I had better change the risk wording for this year's audit.

    More generally, you blog is a great find. I love your fearless analysis.

    As a medical doctor, I am fanatical about patient safety, good design and compliance. I agree, for example, that labeling is pretty useless, but you have to distinguish labeling in UIM, which no one reads, vs on the actual product or in a software menu.

    However, am I the only one seeing an explosion of crazy regulations coming out of EU in the past 2-3 years, which are increasingly poorly defined, confused, contradictory, and sometimes frankly absurd?

    We have had to put on an extra F/T and lost a year in R&D just to handle this red tape. It started with the bizarre 60601 3rd Ed and the NB's and test labs' attempt to hoodwink us into retesting all products (we spent a year retro-complying, without a single safety benefit; I could write a book about our experiences). We pushed back hard and managed to do an internal delta testing against Ed3 with an external review.

    Then there are the endless revisions, changes to AR, local registration, increasing local language requirements even for software GUI for professionals, mandatory EN62304 for NB's but not for MDD, risk management seemingly now risk elimination at all costs, not to mention the raft of runaway environmentalism based on the dubious precautionary principle with WEEE, REACH, Battery, ISO14001 and now RoHS.

    As I see it, the losers will be medical innovation, patient access to affordable medicine and exploding health budgets.

    Perhaps rather than the waterfall model of ever increasing up front efforts at risk and approvals with skyrocketing costs and diminishing returns, and then little after market attention, we need a more agile system, with easier to market, but then a much stronger collaborative and non-punitive after market reporting and correction system. The market might then work more effectively.

    ReplyDelete
  7. Let's have a closer look at labeling: Distinguish between instruction (not only what to do but also what to avoid) and information (of the user on residual risks). Instructions are the third priority means of risk mitigation, in fact lacking in the 2nd Essential Requirement. User information on residual risks is not a means of risk mitigation. With this appraoch I didn't experience any problem with the Notified Body (Class III design dossier).

    ReplyDelete