Coherence between risk analysis and instructions for use in the EU
This article was originally published in RAJ Devices
Adrie de Bruijn and colleagues seek feedback on Dutch studies of a wide variety of medical device technical files that found there was little coherence between risk analysis and instructions for use.
The essential requirements of the European medical device directives (Directive 93/42/EEC on medical devices, Directive 90/385/EEC on active implantable medical devices and Directive 98/79/EC on in vitro diagnostics) require that the solutions adopted by the manufacturer for the design and construction of the devices must conform to safety principles, taking account of the generally acknowledged state of the art1-3. To minimise the risks involved in the use of the device, all known or foreseeable hazards should be identified, risks arising from the identified hazards should be estimated and the actions taken to eliminate or reduce the risks should be clearly described and proven to be adequate4.
In selecting the most appropriate solutions to achieve elimination or reduction of risks, the MDD and the IVD Directive state that the manufacturer must apply the following risk mitigation principles in the following order:
eliminate or reduce risks as far as possible by inherently safe design and construction;
where appropriate, take adequate protection measures including alarms if necessary, in relation to risks that cannot be eliminated; and
inform users of the residual risks due to any shortcomings of the protection measures adopted.
The residual risks should be addressed on the label and in the instructions for use (IFU) as warnings, precautions or contraindications (WPCs), to ensure that the user is properly informed about the risks that are involved with the use of the device and how they can avoid hazardous situations by taking appropriate measures.
In the years 2005-10, the Dutch National Institute for Public Health and the Environment (RIVM) conducted eight studies to evaluate the quality of 114 technical files of a variety of medical devices (see Table 1). Essential items assessed were the risk analysis (RA) and the IFU. We postulate that the quality of these two separate documents can be partly determined by their level of coherence, ie the extent to which residual risks and WPCs are cross-referenced.
Table 1. Technical file evaluations conducted by the RIVM: 2005-2010
Type of devices
Number of files
Annex II medical devices6,13 (ie conformity assessment through Annex II of the MDD)
IIa, IIb, III
Devices for clinical investigation7,14
Over-the-counter medical devices8,15
Ventilators for home use9
Dialysis equipment for home use10
IVDs for self-testing11
Class III medical devices12
The RA and the IFU are controlled in full coherence with one another if: all residual risks that could not be mitigated through design or protection measures including alarms, as analysed in the RA, are actually addressed in the IFU as WPCs; and vice-versa, ie all WPCs that are mentioned in the IFU originate from the risk analysis.
Apart from functioning as a simple tool to reveal omissions in the two documents, the degree of coherence also gives an indication about the extent to which the manufacturer applied the above-mentioned risk mitigation principles in the correct priority order. The risk mitigation principles explicitly mention informing the user about the residual risks as the last option. If a warning, precaution or contraindication is mentioned in the IFU but the corresponding risk is not mentioned in the RA, this risk may not have been properly analysed. As a result, it remains uncertain whether the risk is actually applicable to the device, and if so, whether this risk could have been resolved by adaptation of the design or the addition of a protection measure.
For each of the files that were assessed, the RA was screened for residual risks for which the manufacturer states that a warning, precaution or contraindication should be mentioned in the IFU. Subsequently, the IFU was meticulously verified for the presence of these WPCs. A similar exercise was performed to screen the IFU for WPCs. Subsequently, the risk analysis was checked for the residual risks that justified each of these WPCs.
In general, the coherence assessment showed similar results for each of the eight studies. The overall coherence assessment in the 114 files gave the results as shown in Figure 1. In the majority of the files (63%), most residual risks identified in the RA were mentioned in the IFU. On the other hand, in only a minority of the files (13%) were most WPCs found in the IFU mentioned in the RA. In a small portion of the files (10%) the manufacturer did not identify any residual risks that would necessitate WPCs in the IFU. However, all of the IFUs that we evaluated contained WPCs, even when the manufacturer did not identify any residual risks in the RA.
The lack of coherence between the RA and the IFU we found in the evaluations of 114 technical files raises questions about the thoroughness of the risk mitigation procedure followed by the manufacturers of medical devices and the accuracy of the communication of the residual risks to the user.
The fact that not all of the residual risks that are identified in the RA during the design phase or after the product is placed on the market are found in the IFU as WPCs, could indicate that the RA, or its revisions, are insufficiently used as input for the IFU. It could also indicate that there is a structural lack of communication between several internal processes, including the design process, preparation of information for the users, risk management and post-market surveillance. We recommend that this problem be corrected by the manufacturer by applying stricter procedures for writing IFUs and organising communication between the various processes, eg by linking every residual risk that is identified during the risk analysis to a "to do" list for the people that write the IFU and subsequently checking the coherence between residual risks and WPC in the IFU.
For the presence of WPCs in the IFU that do not originate from the RA, we envisage several possible causes:
the "better-safe-than-sorry" principle: general precautions and warnings such as "Do not use the device for any other purpose than indicated on the label", "Do not modify the device", etc, are written in the IFU for legal reasons (ie to prevent liability claims), although they have no specific bearing on the device in question;
the device has been on the market for many years (eg hip implants) and the WPCs that were put in the original IFU were copied with every revision of the device and/or the IFU, without checking whether the WPCs were still applicable. With a revision of the device, residual risks may actually have been mitigated by modification of the design or protection measures; and
post-marketing surveillance, vigilance and possible other sources of information may have created the need for additional WPCs in the IFU. WPCs are added to the IFU without preceding risk analysis and/or root-cause analysis of the encountered problems. This shows that the continuous cycle of improvement, which implies interaction in both directions between PMS, risk management activities and "design and use information", is not always functioning correctly5.
In a sound risk management system, the proper procedure for modification of the IFU would be such that the new risks that have been identified by PMS and/or vigilance are analysed and that the risk mitigation principles are followed in the correct order. Only if the new risks cannot be mitigated through alteration of the design or the addition of protection measures, should the manufacturer decide whether the risks should be labelled as residual risks and appropriate WPCs added to the IFU. The main objection against short-circuiting these principles by adding additional WPCs in the IFU, without the manufacturer having considered an improved design or additional protective measures for a safer device, is that the user may be exposed to risks that could have been mitigated in other ways. Furthermore, risk analysis is needed to determine whether the risks related to the WPCs are actually applicable to the device. WPCs should be related to the residual risks that are identified in the RA only.
After evaluating 114 technical files of medical devices of various types and within varying risk classes in eight different studies, we consistently found that the necessary coherence between RA and IFU is lacking. One therefore has to ask:
whether the manufacturers' procedures to perform RA and to write IFUs are often not sufficiently robust, especially with regard to communication and interaction between groups responsible for these activities and to the order of the risk mitigation principles as prescribed by the MDD; or
whether our basic assumption – that the level of coherence between the RA and the IFU is an appropriate tool in the assessment of the quality of the RA and IFU documents and an indication for the level at which the risk mitigation principles are implemented – is incorrect.
We would appreciate opinions on this subject. See below for contact details.
1. Directive 93/42/EEC, OJ, 12 July 1993, L169, 1-43, http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CONSLEG:1993L0042:20071011:en:PDF
2. Directive 90/385/EEC, OJ, 20 July 1990, L189, 17-36, http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CONSLEG:1990L0385:20071011:en:PDF
3. Directive 98/79/EEC, OJ, 7 December 1998, L331, 1-37, http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CONSLEG:1998L0079:20031120:en:PDF
4. Medical devices – Application of risk management to medical devices, ISO 14971:2007
5. Roszek B et al, RIVM report 360050014 (2009), Continuous cycle of improvement of medical devices: A questionnaire on experiences and procedures, www.rivm.nl/bibliotheek/rapporten/360050014.pdf
6. Roszek B et al, RIVM report 265011003 (2005), Assessment of technical documentation of Annex II medical devices, www.rivm.nl/bibliotheek/rapporten/265011003.pdf
7. Roszek B et al, RIVM report 360050001 (2006), Assessment of technical documentation of medical devices for clinical investigation, www.rivm.nl/bibliotheek/rapporten/360050001.pdf
8. Hollestelle ML et al, RIVM report 360050002 (2007), Risks associated with the lay use of over the counter medical devices, www.rivm.nl/bibliotheek/rapporten/360050002.pdf
9. Drongelen AW van et al, RIVM report 360050005 (2007), Zijn de risico's van apparatuur voor thuisbeademing door de fabrikanten voldoende afgedekt?, www.rivm.nl/bibliotheek/rapporten/360050005.pdf
10. Vries CGJCA de et al, RIVM report 360050016 (2008), Zijn de risico's van de apparatuur voor thuisdialyse door de fabrikanten voldoende afgedekt?, www.rivm.nl/bibliotheek/rapporten/360050016.pdf
11. Drongelen AW van et al, RIVM report 360050017 (2008), Assessment of files of in vitro diagnostic devices for self-testing, www.rivm.nl/bibliotheek/rapporten/360050017.pdf
12. Roszek B et al, RIVM report 360050021 (2010), Assessment of technical documentation of Class III medical devices, www.rivm.nl/bibliotheek/rapporten/360050021.pdf
13. van Drongelen AW et al, Annex II Technical Documentation Assessed, Med Device Technol, December 2005, 16(10), 21-3
14. Roszek B et al, Improving Technical Documentation for Devices for Clinical Investigation in the EU, Regulatory Affairs Medtech, 20 May 2009
15. Hollestelle M et al, Medical Devices Intended for Lay Use in the EU, Regulatory Affairs Medtech, 18 March 2009
16. de Vries CGJCA et al, RIVM report 360050025 (2010), Point-of-care-devices – An assessment of safety related items in the technical documentation of manufacturers, www.rivm.nl/bibliotheek/rapporten/360050025.pdf
This paper was conceived through the mutual effort of all the following individuals: Adrie de Bruijn, Arjan van Drongelen, Ellen Hilbers, Jacqueline Pot, Boris Roszek, Claudette de Vries and Robert Geertsma, all of whom are scientific officers at the Dutch National Institute for Public Health and the Environment (RIVM), and Jos Kraus, senior inspector of medical technology at the Dutch Health Care Inspectorate (IGZ). Contact email: email@example.com.