Fatal errors in nitrous oxide delivery


Correspondence to: Dr Holger Herff
E-mail: holger.herff@i-med.ac.at


Nitrous oxide continues to be used frequently and the possibility of inadvertent fatal hypoxaemia resulting from technical errors with its administration still exists. A Medline analysis revealed only a few case reports over the last 30 years, and a closed claim analysis only reported ‘claims involving oxygen supply lines’ predating 1990. The aim of this study was to assess the frequency of nitrous oxide-related catastrophes during general anaesthesia in Germany, Austria, and Switzerland. As nitrous oxide-related anaesthesia casualties are rare but generally prosecuted, they almost invariably attract significant media attention. We scanned mass media archives from April 2004 until October 2006 for nitrous oxide-related disasters during general anaesthesia. This approach detected six incidents which were almost certainly nitrous oxide ventilation-related deaths. Searching non-scientific data bases demonstrates that severe incidents involving oxygen supply lines occurred after 1990, and may be much more frequent than previously thought.

Inhalational anaesthesia with nitrous oxide is frequently used. Inadvertent nitrous oxide-mediated complications or even fatal hypoxemia may occur due to technical problems [1] although fortunately, catastrophic nitrous oxide-related complications seem to be rare. An example of such a problem involved an anaesthetist who was called at night to an emergency caesarean section. Oxygen saturation was 92% but following commencement of pre-oxygenation it deteriorated rapidly. On inspection of the anaesthetic machine, the anaesthetist detected that he had inadvertently opened the nitrous oxide valve; due to a technical problem no concurrent oxygen flow was present, resulting in 100% inspiratory nitrous oxide. He then shut down nitrous oxide flow, and switched on oxygen flow; anaesthesia induction was then uneventful, and mother and newborn made a full recovery. In another case a two year-old boy underwent inhalational induction for a CT scan of the head with 50% oxygen, 50% nitrous oxide, and sevoflurane. The child then immediately started to hyperventilate, while oxygen saturation decreased rapidly to 80%. The display of the anaesthesia machine indicated ‘O2 failure’. Due to frequent re-positioning of anaesthesia equipment in the small radiology suite, disconnection of oxygen pipeline in the wall socket was immediately suspected, identified, and corrected. Obviously in both cases the mechanical interlock between oxygen and nitrous oxide flow meter controls failed, resulting in a hypoxic ventilation gas with high inspired nitrous oxide concentration, a failure with the potential to harm patients severely and which has been reported before [2].

Interestingly, Medline analysis targeting severe nitrous oxide-related incidents revealed only a few case reports in the last three decades [3, 4]. Further, a closed claim analysis reported only ‘claims involving oxygen supply lines’ predating 1990 [5], or a general decrease of respiratory-related damaging events [6]. The aim of this study was to assess the frequency of nitrous oxide-related catastrophes during general anaesthesia in Germany, Austria, and Switzerland (total population, ∼ 100 million).


We were not able to detect recent reports in a Medline analysis concerning fatal nitrous oxide-related anaesthesia complications. As these are rare but usually result in legal action, they attract significant media attention. Thus, we scanned mass media archives over a two year period from April 2004 until October 2006 for nitrous oxide-related disasters during general anaesthesia. For practical and linguistic reasons this was limited to the three German-speaking countries: Germany, Austria and Switzerland. A total of 30 newspapers, three weekly magazines, and three national TV-channel internet-based news archives (Table 1) were searched for the term ‘Lachgas’ (nitrous oxide). As all three countries are federal states creating numerous daily regional newspapers, we attempted to evaluate archives from every major region (Fig. 1). Further, we utilised a special ‘search-engine’ for print-media (‘Genios’ media archive), including archives of an additional 110 local newspapers and magazines in the three countries, for the term ‘Lachgas’. The archives provided full text search for the search term and as results delivered headlines, and small summaries of the articles. We then purchased the whole article if it appeared to provide new or exclusive information. The information presented here includes details of the investigation, accusations made by the state attorney or the final legal outcome based on expert opinion. The emphasis of our analysis was identification of a single event, followed by an attempt to obtain details of the incident, its mechanism and the circumstances of its development, and finally, if the case was legally closed, whether cause was proven. Although details of the location, individuals and institutions involved were mentioned in media reports, these details have not been included in this paper.

Table 1.   Newspapers, magazines and TV-channels/internet archives included in the media search.
CountryDaily Newspapers MagazinesTV Station/Internet
GermanyBerliner ZeitungBonner GeneralanzeigerDer SpiegelARD/Tagesschau
Frankfurter RundschauFreies Wort/SuhlFocusSat1 News
Hamburger MorgenpostKieler NachrichtenSternZDF/Heute
Mitteldeutsche Zeitung
Nordwest Zeitung
Passauer Neue Presse
 Genios-archive (including > 140
 newspapers, magazines, etc.)
Rheinische PostRostocker Zeitung  
Saechsische ZeitungSaarbruecker Zeitung  
Schweriner VolkszeitungStuttgarter Nachrichten  
Westdeutsche Allgemeine ZeitungWestfaelische Zeitung  
AustriaDie PresseKleine Zeitung  
KurierOberoesterreichische Nachrichten  
Salzburger NachrichtenStandard  
Tiroler Tageszeitung   
SwitzerlandBasler ZeitungBerner Zeitung  
Neue Zuercher Zeitung   
Figure 1.

 Map of the head offices of media-archives included in the study. ▪ Capitals and ◆ major cities are marked to facilitate orientation. (inline image) Head office of a daily newspaper (only one is marked per city, even if there are more). (+) Head office of a weekly magazine. (inline image) Headquarters of a TV Station.


With the general search term ‘Lachgas’ there were numerous articles identified concerning other uses of nitrous oxide as, e.g. drug abuse, its role in climate changes, or even its use in car tuning. We identified six cases with a fatal outcome in anaesthetic practice and these were most probably due to ventilation with excessive nitrous oxide (Table 2). We did not identify any cases resulting in non-fatal morbidity. A cross-check employing a ‘search-engine’ of 110 additional print-media sources confirmed these six cases, but no further incidents. All cases were first reported over a two year period between April 2004 and March 2006.

Table 2.   Search-hits for articles reporting of nitrous oxide-related deaths during anaesthesia in newspapers, magazines or TV stations.
CasesSource of PublicationDate of publication
Case 1 (Austria; 40 years; spine surgery)Salzburger Nachrichten24 October 2005
Standard30 October 2005
Tiroler Tageszeitungmultiple, e.g. 25 October 2005
Kleine Zeitung19 September 2006
Case 2 (Germany; 19 years; post-traumatic surgery)Sueddeutsche Zeitung26 & 27 March 2006
Case 3 (Switzerland; 58 years; cardiac surgery)Neue Zuercher Zeitung4 April 2004
Cases 4–6 (Germany; 3 women for caesarean section in one hospital)ARD/Tagesschau24 December 2004
Focus10 July 2006
Frankfurter Rundschau4 September 2004
Saechsische Zeitungmutiple, e.g. 21 December 2005
Sueddeutsche Zeitung10 September 2004

Case 1

In a newly opened outpatient surgical centre in Austria, the first anaesthetised patient developed hypoxia during induction of anaesthesia, and died after unsuccessful cardiopulmonary resuscitation (CPR) efforts. The state attorney accused a technician of connecting the oxygen pipeline with a nitrous oxide tank in the construction phase, and blamed the anaesthesiologist for not discovering the error.

Case 2

In Germany, a 19-year-old man had an accident with his motorbike but was obviously only slightly injured. During induction of anaesthesia, the patient developed cardiac arrest and died despite CPR efforts. Investigations by the state attorney revealed that the oxygen pipeline was connected with the nitrous oxide tank in the basement of the hospital.

Case 3

In Switzerland, the oxygen pipeline of a cardiopulmonary bypass machine was connected to the nitrous oxide supply; the patient died in the operating room due to hypoxia. The ‘Neue Zuercher Zeitung’, a Swiss national newspaper, reported on 4 April 2004 that the oxygen pipeline had been fitted to the nitrous oxide wall sockets of the central gas supply. It pointed out that if previous nitrous oxide misconnections had been reported to a critical incident report system this might have helped to avoid this fatal error [7, 8].

Cases 4–6

In Germany, four women died within several weeks in one hospital when undergoing induction of anaesthesia for caesarean section. In three cases, the state attorney proved that the oxygen and nitrous oxide pipeline were misconnected within the anaesthesia machine. The responsible technician was formally sentenced.


Annually, about 4.7 million general anaesthesia cases are performed in Germany (Report of the 47th German Anaesthesia Congress, Sueddeutsche Zeitung; 6 May 2000). Extrapolating this number to the German-speaking countries Germany, Austria, and Switzerland, would indicate approximately 11 million anaesthesia cases within two years. Exact data of anaesthesia related deaths from these countries are lacking. Extrapolating international incidence rates of anaesthesia-related casualties to the 11 million anaesthesia cases, we might expect 70–100 deaths within two years [8–14].

There are multiple reasons for cardiac arrest during anaesthesia, with airway-related disasters constituting at least 50% of anaesthesia-related deaths [8, 13, 15]. Less likely fatal complications include arrhythmias [14], anaphylactic reactions [15], and malignant hyperthermia [16]. Based on our research, six of 70–100 anaesthesia-related deaths were probably due to inadvertent ventilation with excessive nitrous oxide. Only a few case reports describing this phenomenon were identified over 30 years [2, 3], and closed claims analyses or mortality data bases, which were not targeting this phenomenon explicitly, revealed no current cases [4, 5, 12]. Thus, either there are no cases, or these scientific tools may have failed to detect or have underestimated the number of nitrous oxide-related casualties. In addition, there were no alerts by the existing local voluntary incident report systems. However, it is most unlikely that any physician accused of negligent homicide will report any details to a critical incident reporting system. This failure of reporting of clinical disasters render clinical ‘routine incident reporting systems [to] be poor at identifying patient safety incidents, particularly those resulting in harm’[17]. With scientific tools and incident report systems obviously failing, the actual incidence of nitrous oxide-related casualties may have been underestimated. In the future, more awareness within the medical community to detect rare but fatal events is needed; the lay press discussed these cases intensely and the medical community should do so as well.

In contrast to the reports in the introduction with a failing mechanical interlock between oxygen and nitrous oxide flow meter controls resulting in a hypoxic ventilation gas, all six incidents were most probably caused by misconnected oxygen and nitrous oxide pipelines. Human errors account for the vast majority (82%) of adverse events during anaesthesia, with technical failure accounting for only 14%[18]. The human error of misconnecting oxygen pipelines with a nitrous oxide-tank, which may not even be located in the anaesthesia machine or in the operating room itself, results in technical failure which may subsequently not be recognised due to a further human error. This chain of errors may make it extremely difficult to detect a misconnection during anaesthesia [19]. Technical failure in gas supply is so rare it may be as unthinkable as water in the fuel tank of a car. In a simulator scenario in New Zealand, all 20 anaesthetists studied were able to cope safely with a central oxygen supply failure due to total loss of gas pressure [20]. Nevertheless, ‘no anaesthetist disconnected the wall pipeline supply, and all subsequently used the reconnected oxygen pipeline supply without testing, with the potential to deliver contaminated gases to the patient’[20], demonstrating the anaesthetist's trust in the gases delivered by central gas supply.

Misconnection of the oxygen pipeline may be very difficult to recognise if monitoring of the oxygen concentration in the inspired gas mixture is not undertaken, particularly when nitrous oxide-mediated hypoxia with cardiovascular collapse is developing, and then imminent. Although the responsible anaesthesiologist must check the anaesthesia machine every time before induction of anaesthesia [21], it is prudent to treat an anaesthesia machine especially carefully after repair or service to avoid nitrous oxide-related catastrophes [22].

Potent opioids such as remifentanil have reduced the need for intra-operative nitrous oxide analgesia, and the use of nitrous oxide is decreasing [23]. Some hospitals have also turned off the nitrous oxide gas supply or even removed all nitrous oxide supply pipelines, which may be a strategy to avoid further nitrous oxide-related catastrophes [24, 25]. According to European Union law [26] standards for anaesthesia machines have been set by the European Commission (Anaesthetic Workstations and their modules – Particular Requirements; EN 740 : 1998/A1: 2004) including different safety mechanisms to prevent misconnection of gas supply pipelines or excessive nitrous oxide ventilation (Switzerland is not an EU member, but has implemented EU standards for medical products in national law: Medizinprodukte Verordnung; MepV: SR 812.213). Among these prevention standards is the mechanical interlock between oxygen and nitrous oxide flow meter controls, which may help to avoid hypoxaemia if all connections are fitted correctly; a fail-safe mechanism for oxygen supply line failure may warn of disconnection. Further, pin indexing for medical gases should avoid misconnection in daily routine use. In our study, misconnection obviously occurred when pipelines were reconnected after technical service or repair, or in the construction phase. Thus, it is unlikely that such a misconnection could be avoided with standard safety mechanisms. The only technical mechanism to detect misconnected pipelines and nitrous oxide coming out of the oxygen pipeline may be an oxygen analyser. Therefore, according to European standards (EN 740) oxygen analysers must be incorporated in all anaesthesia machines and according to American Society of Anesthesiologists (ASA) and Association of Anaesthetists of Great Britain and Ireland (AAGBI) standards, for example, no anaesthetic should be performed without them [27]. Furthermore, careful pre-oxygenation of each patient may render the patient an indirect controller of his anaesthesia machine. An increase in oxygen saturation while a high FIo2 is indicated suggests that the anaesthesia machine is working properly [28, 29]. In contrast, falling oxygen saturation during pre-oxygenation may indicate insufflation of an hypoxic gas mixture.

We were unable to obtain exact reports of the incidents other than media coverage. Legal negotiations are still ongoing in two of the six cases, thus presumption of innocence must be maintained for those accused. We may have missed more nitrous oxide-related deaths, or these six cases may be simply a coincidental accumulation of nitrous oxide-related catastrophes in the observation period. Further, 11 million anaesthesia cases within two years is only an estimate. With reliable denominator data lacking, no incidence rates or epidemiological indicators could be calculated from this study. We cannot report about less severe nitrous oxide-related incidents, which would probably not attract media attention. Finally, this may not be representative for other countries. Nevertheless, this approach detected six anaesthesia-related deaths by searching non-scientific data bases [30], and demonstrates that severe incidents involving oxygen supply lines did occur after 1990, despite assumptions to the contrary [5] and may be more frequent in these countries than previously thought.


Supported, in part, by the Science Foundation of the Austrian National Bank, grant 11448, Vienna, Austria.

No author has any conflict of interest in regard of devices or methods described in this article.

As no direct patient or other medical data were used for this study, no ethics approval was required.