Analysis of the French health ministry’s national register of incidents involving medical devices in anaesthesia and intensive care

L. Beydon1, F. Conreux1, R. Le Gall1, D. Safran2, J. B. Cazalaa3 and the members of the ‘Sous-commission de Materiovigilance’ for Anaesthesia and Intensive Care,{dagger}

1Département d’Anaesthésie, CHU d’Angers, F-49033 Angers Cedex 01, France. 2Département d’Anaesthésie, Hôpital Necker, F-75015 Paris et Commission Nationale de Matériovigilance, France 3Département d’Anaesthésie, Hôpital Laennec, Rue de Sèvres, F-75007 Paris, France*Corresponding author

{dagger}J. Ancellin, B. Bacchi, J. M. Bedicam, V. Billard, M. A. Dauphin, S. Fougère, G. Freys, F. Fries, C. Itty, G. Krim, J. Labrousse, G. Laguenie, R. Leroyer, P. Lever, M. A. Lochet, M. P. Macquet, D. Meunier, S. Molliex, J. F. Muir, Y. Nivoche, J. C. Otteni, E. Salgues, F. Saulnier, D. Thiveaud and J. P. Viale

Accepted for publication: November 7, 2000


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
This study details all incidents involving medical devices used in anaesthesia and intensive care reported to the relevant authorities in France in 1998. There were 1004 reports during that year. Incidents were classified as serious (harmful to patients) in 11% of cases; death resulted in 2% of cases. Equipment for ventilation and infusion, and monitors of all kinds, accounted for most of the reports, representing 37%, 30% and 12%, respectively, of all reports. The leading causes of failure varied according to the category of device. User errors, quality control problems during production of the device and design faults were the three main causes. The problems identified during the study period enabled the faulty medical devices to be improved in 12–44% of cases. We conclude that post-marketing vigilance is a useful way of improving the quality of medical devices.

Br J Anaesth 2001; 86: 382–7

Keywords: equipment, anaesthesia machines; equipment, failure; monitoring, intensive care


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Most accidents in anaesthesia are caused by human error,1 2 but medical devices may also fail and do cause critical incidents. Such events have appeared in the medical literature since the earliest days of case reports, but most are anecdotal.3 The traditional approach in this field has been to collect incident reports in a single clinical setting, e.g. ICU or anaesthesia, and to list all the clinical mishaps that have occurred in a certain period. The aim has been to assess mishaps as a whole. Pioneering studies by Cooper, Newbower and Kitz used closed-claims analysis to track a large series of incidents.4 An Australian prospective analysis of 2000 incidents in anaesthesia represents another interesting approach.5 6 However, of all the factors studied, devices have been involved only marginally, and very few series have focused on them.4,7 Post-marketing surveillance and reporting systems have opened a new era in the collection of device-related mishaps. Incident reporting managed by the United States Federal Drug Administration (FDA) started in 1974. Under European Community regulations, many European countries also now have this kind of reporting. The principle has been applied by law in France since 1996,8 and details have been published.9 10 Under this decree, anyone who is aware of an incident in which a patient has been or may have been injured must notify the Agence Française de Sécurité Sanitaire des Produits de Santé (AFSSAPS). The reporting of less serious incidents involving medical devices is also being encouraged. Despite the extension of post-marketing surveillance and reporting to all European countries, no data have yet been published. We aim to begin this process by describing key findings extracted from the French national database of 1998 post-marketing reports in anaesthesia and intensive care.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
We have analysed all the incident reports in anaesthesia and intensive care medicine which were sent to and processed by the French authorities in 1998. These reports concerned medical devices that had failed or which injured or may have injured patients. As the launch of the post-marketing process was still relatively recent in France, voluntary reports of less severe incidents were also encouraged. Incident reports were sent by public and private hospitals and by health organizations from other European countries. Although manufacturers are required by European regulation and by their quality assurance policy to notify problems with their products, few such notifications were received. Each report was set out on a pre-printed form designed by the French authorities. This listed the type of device, a short description of the incident and the name of the reporting institution. Whenever it was judged necessary, the authorities contacted the institution submitting the report for additional details and prompted the manufacturer to examine the device. In complex cases, a clinical or bioengineering expert visited the centre to see the device and review the facts with those who had made the original report. In a very few cases, an independent laboratory was asked to examine the device.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
We analysed the 1004 reports listed in the database in 1998. At the time we examined them, all reports had been finalized. The reports were divided into nine categories (Table 1). ‘Regional’ anaesthesia refers to all single-use needles and catheters used in this kind of anaesthesia. ‘Catheter ports’ refers to the type of implanted device (including a chamber and a catheter) mostly used for the perfusion of cytotoxic drugs. In France, these devices belong to the field of intensive care. The ‘infusion’ category includes catheters (46%), syringes and tubing (10%), syringe pumps (23%) and all other forms of needles, syringes, filters and other single-use equipment (21%). The ‘defibrillator’ category includes paddles, battery chargers and other electrical supplies. ‘Medical gas delivery’ refers to delivery systems for oxygen, nitrous oxide, air and vacuum (from the tank through to the wall of the room where the gas is used). ‘Incubators’ are those in neonatal intensive care. The ‘monitors’ category includes all kinds of device used in intensive care or the operating theatre whether compact (19%) or modular (55%), accessories such as SpO2 sensors (8%) and all other kinds of monitor (18%). The ‘rewarming devices’ category includes those used to rewarm patients (excluding incubators) or infusion fluids. It also includes pumps used to accelerate transfusion. ‘Ventilation’ refers to equipment used to provide supplemental oxygen or to ventilate patients’ lungs, including masks, tracheal tubes (25%), ventilators of any type (51%) and attached components, e.g. unidirectional valves (4%), circuits (6%), volatile anaesthetic vaporizers (2%), gas mixers (3%) and accessories (9%).


View this table:
[in this window]
[in a new window]
 
Table 1 Incidents by category of medical device (MD). *% of declarations for each class of MD; regional=regional anaesthesia; rewarming=devices used to warm patients or perfusion fluids
 
About 30% of the incidents were reported by users to be severe (an incident which harmed or may have harmed the patient) (Table 1). However, after review by a member of the French authorities, only 11% retained this rating.

Twenty reports (2%) were associated with patient death. The failure of the device itself accounted for 12 accidents: two ventilators of the same type caused massive pneumothorax when they failed; this led to changes in the design of the apparatus. The devices involved in all the other cases of device failure were different in each case. In four cases the monitors did not sound an alarm in response to ventricular fibrillation or asystole. In two other cases, the monitors mistook the activity of a pacemaker for that of the heart, and produced no alarm even when the patient was already dead. Two cases of gas supply failure each killed one patient through contamination of the oxygen line with nitrous oxide. In two cases an isoflurane vaporizer gave an overdose, resulting in the death of one patient. Incorrect use was the second most common cause of death (n=7). One user failed to make proper use of an oxygen tank for home ventilation and died from anoxia. Another death was attributed to a heat and moisture exchanger (HME), which was attached to a one-way Ambu valve in the wrong direction; ventilation became impossible and the patient died from anoxia. This problem with one-way valves prompted an order from the French authorities for design changes to make such mistakes impossible. The four deaths involving infusion equipment were associated with catheter fracture and migration, cardiac tamponade and embolism, usually resulting from technical errors made during insertion of catheters. One patient died because all alarms had been turned off to make life more comfortable for the nursing staff. We encountered one death in which the device did not seem to be faulty. The clinician using a pulse oximeter failed to identify as ‘false’ a very low SpO2 value during extracorporeal circulation for repair of a congenital defect, although pulse oximetry is known to be unreliable in this context. However, this problem was not the main cause of death.

While 98% of the accidents were not fatal, several of them warrant further examination. Some of these cases involve the same problems as the fatal ones described above.

The rupture of epidural catheters in the epidural space was the most significant problem in regional anaesthesia. The most significant catheter port incident was catheter fracture (44% of reports for catheter ports), leading to their migration into the heart and pulmonary veins in 36% of cases.

Defibrillation equipment caused burns to patient or doctor in three cases (11%). A shock could not be delivered in four cases. Three cases were the result of incorrect use. For example, synchronization to the ECG requires a delay before the shock can be triggered and delivered. This function was not understood by the user, who thought the device had failed. Other problems included the unexpected failure of batteries or components. Checking procedures had often not been performed properly or at the correct time.

Incidents involving systems for medical gas delivery and gas analysers were mostly minor, such as leaks, simple failure of gas analysers or rotameter inaccuracies. However, serious errors involved the misconnection of medical gas lines; these occurred in four cases and were possibly caused by defects in the sockets. Still more seriously, there were four cases of ignition of aluminum oxygen pressure regulators after the oxygen high pressure valve, upstream of the regulator, was opened rapidly, causing major burns to users. Two non-fatal cases of contamination of the oxygen lines with nitrous oxide were noted.

For infusion equipment (excluding catheter ports) there were few severe accidents. These included the migration of catheters inside vessels, air embolism and infusion rate errors with syringe pumps. Minor problems were numerous and mainly caused by errors of use and/or failures of quality assurance during manufacture.

The extent of problems with monitors varied according to the type. Haemodynamic monitors (45%) headed the list, followed by SpO2 monitors, including their accessories (15%). Monitors failed mainly because of component breakdown and errors in design, including software (Table 3).


View this table:
[in this window]
[in a new window]
 
Table 3 Problems identified, and cases where manufacturer modified the design of the medical device. Percentages do not always add up to 100% for each class of device. In some cases two causes were found; in others none was identified. *% of declarations for each class of device; regional=regional anaesthesia; rewarming=devices used to warm patients or perfusion fluids; {dagger}devices were declared obsolete according to French or European standards
 
Several cases involved problems of ventilation. Unexpected failure without an alarm occurred with 23 ventilators, and barotrauma with seven. Disposable one-way valves were reversed when an HME filter was placed between the valve and Y-piece in three non-fatal cases. A tracheostomy tube migrated into one patient’s trachea; this was caused by the connecting piece becoming detatched. Enquiries showed that the glue used in the production of these devices could melt when iodine and alcoholic solutions were used to clean the tracheostomy incision. Failures of tracheal tube securing devices were reported in five instances, allowing the tube to twist and reduce ventilation. A design fault seemed to be the cause in most cases.

The database was then analysed to assess the mode of investigation used, and the causes of failure, classified by category of equipment (Tables 2 and 3). Table 2 shows the number of reports that required a formal analysis of the faulty device by the manufacturer or an independent laboratory. Such a formal analysis was possible in only 40% of cases involving perfusion catheters, catheter ports and equipment for regional anaesthesia. This was because most users, having declared the problem, discarded the device, either by accident or on the grounds that it was ‘contaminated with blood’ and thus not safe to send to a laboratory. Thus, effective laboratory investigations were performed in fewer than 33% of cases. This was due to the large number of component breakdowns that were easily recognized without specific expertise.


View this table:
[in this window]
[in a new window]
 
Table 2 Nature of inquiry for each category of medical device. *% of declarations for each class of device; regional=regional anaesthesia; rewarming=devices used to warm patients or perfusion fluids
 
The origin of the problem, whenever this could be identified, is described in Table 3. Incorrect use was a factor in >70% of the cases involving catheter ports and equipment for regional anaesthesia. Incubators were judged to be obsolete, according to recent French norms, in 20% of cases. Maintenance was found to be a significant cause of failure only for medical gas supplies. Through quality assurance failures or design faults, including software, manufacturers could be judged responsible for the problems in 30–40% of cases. Simple component failure was more evenly distributed among the various categories of device.

Considering the whole database, it is clear that this post-marketing survey led to design changes in 12–44% of cases. These reports led the French health authorities to issue 11 directives demanding the withdrawal from the market or restricted use of the problematic devices. In almost all cases there was a design fault.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The French national reporting system is very similar to those of other European countries11 and North America.12 Its object is to maintain patient safety. Devices are controlled by the ISO9000 series of quality system standards, which apply to manufacturers; these aim for high quality and safety throughout the life of the product from its design through to its use and maintenance. The incident reporting principle complements the ISO9000 system, as it allows users to report problems that might have escaped the manufacturer or which may appear after the release of the device on to the market. This study is the first comprehensive analysis of French post-marketing activity since it began in the field of anaesthesia and intensive care in 1996.

Vigilance is important in anaesthesia as failure of devices used in this field may place the patient at high risk.13 Hart and colleagues6 categorized 22% of incidents reported in ICUs as severe (this category included those in which patients died and other severe incidents). The duration of injuries was permanent or long-lasting in 6%. Equipment featured in 41% of cases. Gilron14 in Canada noted that, although anaesthetic devices accounted for only 2.3% of newly marketed medical devices, they accounted for 8.6% of all problems and recalls and 37% of alerts. Similarly, anaesthesia accounted for 32% of all reports in France.

Infusion equipment and catheter ports
All the incidents we encountered in this field involved known problems. Errors during insertion and failure to check the catheter position radiologically accounted for the majority of problems with catheters. This led to vein perforation and to catheter rupture with migration and leaks. Catheter ports failed and migrated because of unnoticed ‘pinch-offs’.15 16 Like others, we found many errors in the use of infusion pumps.17 This highlights the fact that further education and safety measures are required, even with simple and commonly used devices such as catheters and pumps.

Monitors
Monitors, especially pulse oximeters and capnometers, are important for detecting anaesthetic mishaps, so increasing their reliability is a key issue.18 19 Short and colleagues20 noted that, of the 125 critical incidents reported during a quality assurance programme, 50% were detected by the anesthetist and 34% by monitoring equipment. This highlights the potential risk posed by monitor failures, especially if there is no built-in alarm system.

Defibrillators
Defibrillator failure because of defective batteries is the major cause of failure in published reports.21 The systematic use of checklists and tests has been advocated in response to incidents, especially for defibrillators that are seldom used. Errors of operation were also frequent, especially in devices linked to the patient’s ECG. Users did not know how these differed in function from more classical devices. This highlights a need to educate users and to label the front panel clearly with instructions for fast, error-free operation.

Ventilation and medical gas supply
In most cases, problems arising from ventilation devices were caused by device failure or human error. There were two main problems with medical gas supply: (i) contamination of oxygen lines with nitrous oxide or air due to a valve failure2224 and (ii) ignition of oxygen flow regulators made of aluminium alloy, which can melt at a low temperature, causing serious burns to users. These devices are liable to blow their fuses, which may ignite any inflammable equipment. This has encouraged the French authorities to forbid the use and sale of regulators including parts made of aluminium.

Incubators
Several devices were not sufficiently protected against possible errors of use. Serious burns arose from errors in incubator settings, or in placing the temperature probe on the child. Another was caused by incorrectly placed surgical drapes directing a hot air stream towards the baby. This risk was mentioned in the user manual, but the manufacturers had not gone on to reinforce it with an alarm. A warning from health authorities, an information campaign and the addition of warning labels on the canopy did not prevent another accident of the same type from occurring a year later. This highlights the fact that manufacturers should consider how people actually use their devices in practice, e.g. performing surgical operations in closed incubators instead of open ones.

Human errors were very frequent: they involved 14–71% of the reports, depending on the category of the device. Most of these incidents were caused by negligence or inadequate knowledge, despite clear instruction manuals. It seems that medical device-orientated ‘culture’ is not sufficiently widespread and does not keep up with industrial progress. Williamson and colleagues25 reported in the Australian Incident Monitoring Study that errors of judgement when using medical devices, a failure to check and technical errors represented 16%, 13% and 13% of human errors, respectively. Clayton26 showed that 55% of residents were not able to check anaesthesia machines according to the standards of the New Zealand College of Anaesthetists and that only 16% of them could use defibrillators properly and solve problems involving this device. Reducing errors of this kind will require more education. For this reason, the German authorities have requested users to undergo specific training on each new medical device introduced in a clinical department before they use them on patients.

Another study identified a failure of regular checking in 33% of the reports.21 This was taken up as a major regulatory issue by the French authorities, who require by law that all devices used in the operating theatre should be checked daily before use and according to formal checklists. This recommendation also imposes a ‘between case’ simplified check.27

We found incidents of equipment failure that differed from those reported by Webb and colleagues5 Equipment failures represented only 9% of all incidents reported in their study. Ventilation equipment accounted for 44%. These were due to one-way valve failures in 43% of cases and to leaks of various kinds in 13%. The remaining causes were miscellaneous. Forty per cent of incidents involving ventilators were actually or potentially life-threatening. Monitors were involved in 24% of incidents, of which arterial pressure monitors accounted for 26%. All were described as potentially life-threatening. Oximeters, oxygen analysers, capnographs and pulse oximeters each accounted for about 20%. Surprisingly, infusion pumps accounted for fewer than 2% of all incidents. This pattern of incidents is strikingly different from ours. It may be that the types of event reported in the two studies differed. In the study of Webb and colleagues,5 all consecutive events occurring during anaesthesia were reported, including malpractice and probably also numerous trivial problems resulting from human error. Under these circumstances, equipment failures were likely to be underestimated. Conversely, our data featured equipment failures specifically. However, both studies relied on voluntary reporting. This underestimates the true occurrence of incidents. For example, a study by Sandborn and colleagues28 showed that only 4% of electronically detected incidents had led to a voluntary report. To exploit equipment failures in improving safety and reliability, and to further the education of users, incident reporting must become mandatory. Since ECRI (Emergency Care Research Institute, 19462-1298 Plymouth Meeting, Pennsylvania, USA) and a systematic survey and analysis of medical device failures was introduced in the USA, this reporting has been shown to be successful.29

We conclude that vigilance leads to a better understanding of the problems involving medical devices. The French experience is positive, and may help to improve safety.


    Acknowledgements
 
The authors thank all the members of AFSSAPS who processed reports for their valuable help in data analysis. The English text was edited by G. Watts and Owen Parkes.


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
1 Cooper JB, Newbower RS, Long CD, McPeek B. Preventable anesthesia mishaps: a study of human factors. Anesthesiology 1978; 49: 399–406[ISI][Medline]

2 Wright D, Mackenzie SJ, Buchan I, Cairns CS, Price LE. Critical incidents in the intensive therapy unit. Lancet 1991; 338: 676–8[ISI][Medline]

3 Blum LL. Equipment design and ‘human limitations’. Anasthesiology 1971; 35: 101–2

4 Cooper JB, Newbower RS, Kitz RJ. An analysis of major errors and equipment failures in anesthesia management: considerations for prevention and detection. Anesthesiology 1984; 60: 34–42[ISI][Medline]

5 Webb RK, Russel WJ, Klepper I, Runciman WB. Equipment failure: an analysis of 2000 incident reports. Anaesth Intensive Care 1993; 21: 673–7[ISI][Medline]

6 Hart GK, Baldwin I, Gutteridge G, Ford J. Adverse incident reporting in intensive care. Anaesth Intensive Care 1993; 22: 556–61[ISI]

7 Cheney FW, Posner KL, Caplan RA, Gild WM. Burns from warming devices in anesthesia. A closed claim study. Anesthesiology 1994; 80: 806–10[ISI][Medline]

8 Ministère de la Santé et des Affaires Sociales. Décret 96.32 du 15 janvier 1996, relatif à la materiovigilance exercée sur les dispositifs médicaux. J Officiel République Française 1996; 803–5

9 Cazlàa JB. La matériovigilance à l’échelon central. Bilan et perspectives. Ann Fr Anaesth Réanim 1999; 18: 249–54[ISI][Medline]

10 Freeman MF. Medical device vigilance: the European approach. Stud Health Technol Inform 1996; 28: 13–16[Medline]

11 Ecker W, De Roo R, Frellsen M, Gomes AF, Gutowski R. Activities in other European countries. Stud Health Technol Inform 1996; 28: 119–23[Medline]

12 Gross TP, Kessler LG. Medical device vigilance at FDA. Stud Health Technol Inform 1996; 28: 17–24[Medline]

13 Moussavi S, Whitmore K. Review of scheduled performance assurance inspections. J Clin Eng 1993; 18: 159–64[Medline]

14 Gilron I. Anaesthesia equipment safety in Canada: the role of government regulation. Can J Anaesth 1993; 40: 987–92[Abstract]

15 Andris DA, Krzywda EA, Schulte W, Ausman R, Quebbeman EJ. Pinch-off syndrome: a rare etiology for central venous catheter occlusion. JPEN 1994; 18: 531–3[Abstract]

16 Desruennes E. Mechanical complications at implantation sites. Pathol Biol (Paris) 1999; 47: 269–72[ISI][Medline]

17 Brown SL, Morrison AE, Parmentier CM, Woo EK, Vishnuvajjala RL. Infusion pump adverse events: experience from medical device reports. J Intraven Nurs 1997; 20: 41–9[Medline]

18 Tinker JH, Dull DL, Caplan RA, Ward RJ, Cheney FW. Role of monitoring devices in prevention of anaesthetic mishaps: a closed claims analysis. Anaesthesiology 1989; 71: 541–6

19 Webb RK, Van Der Walt JH, Runciman WB et al. Which monitor? An analysis of 2000 incident reports. Anaesth Intensive Care 1993; 21: 529–42[ISI][Medline]

20 Short TG, O’Regan A, Lew J, Oh TE. Critical incident reporting in an anaesthetic department quality assurance programme. Anaesthesia 1993; 48: 3–7[ISI][Medline]

21 White RD. Maintenance of defibrillators in a state of readiness. Ann Emerg Med 1993; 22: 302–6[ISI][Medline]

22 Mazze RI. Therapeutic misadventure with oxygen delivery systems: the need for continuous in-line oxygen monitors. Anesth Analg 1978; 51: 787–792[Medline]

23 Lye A, Patrick R. Oxygen contamination of the nitrous oxide pipeline supply. Anaesth Intensive Care 1998; 26: 207–9[ISI][Medline]

24 Otteni JC, Ancellin J, Cazalàa JB, Clergue F. Pollution et retropollution du réseau de distribution d’un gaz médical. Ann Fr Anaesth Réanim 1994; 13: 713–25[ISI][Medline]

25 Williamson JA, Webb RK, Sellen A, Runciman WB, Van Der Walt JH. Human failure: an analysis of 2000 incident reports. Anaesth Intensive Care 1993; 21: 678–83[ISI][Medline]

26 Clayton Petty W. The aging anaesthetic machine. J Clin Monit 1997; 13: 129[Medline]

27 Ministère de la Santé et des Affaires Sociales. Décret du 3 octobre 1995 relatif aux modalités d’utilisation et de contrôle des matériels et dispositifs médicaux assurant les fonctions et actes cités aux articles D712–43 et D712–47 du code de la santé publique. J Officiel République Française 1995; 14932–3

28 Sanborn KV, Castro J, Kuroda M, Thys DM. Detection of intraoperative incidents by electronic scanning of computerized anesthesia records. Comparison with voluntary reporting. Anesthesiology 1996; 85: 977–87[ISI][Medline]

29 Nobel JJ. Medical device accident reporting: does it improve patient safety? Stud Health Technol Inform 1996; 28: 29–35[Medline]





This Article
Abstract
Full Text (PDF)
E-Letters: Submit a response to the article
Alert me when this article is cited
Alert me when E-letters are posted
Alert me if a correction is posted
Services
Email this article to a friend
Similar articles in this journal
Similar articles in ISI Web of Science
Similar articles in PubMed
Alert me to new issues of the journal
Add to My Personal Archive
Download to citation manager
Search for citing articles in:
ISI Web of Science (7)
Disclaimer
Request Permissions
Google Scholar
Articles by Beydon, L.
Articles by Cazalaa, J. B.
PubMed
PubMed Citation
Articles by Beydon, L.
Articles by Cazalaa, J. B.