ALL halogenated anesthetics are known to undergo degradation in the presence of strong alkali. Examples of the byproducts of degradation include 2-bromo-2-chloro-1,1-difluoroethene from the haloalkane halothane, carbon monoxide from the difluoromethyl ethers enflurane, isoflurane, and desflurane, and compound A from the monofluoromethyl ether sevoflurane.1These degradation reactions are exothermic, with those involving sevoflurane being particularly energetic.2–7Under laboratory conditions, degradation reactions involving sevoflurane and barium hydroxide lime have reached temperatures in excess of 350–400°C and melting of plastic components, smoldering, and fire have been reported.2,3,5We wish to report an explosion and fire in a carbon dioxide absorber containing barium hydroxide lime exposed to oxygen and sevoflurane.
At the Cincinnati Children’s Hospital Medical Center many operating rooms have a small (approximately 12 m2), adjacent induction room with a separate anesthesia machine; most inductions take place in the induction room in the presence of a parent. Inductions are typically performed by inhalation of sevoflurane in nitrous oxide and oxygen at a total fresh gas flow of 10–12 l/min. Once the patient has reached an appropriate level of anesthesia, the parent is escorted to the waiting room and the patient is transferred through a door into the operating room. The anesthesia circuit and bag are removed from the induction room anesthesia machine and transferred to the operating room with the patient. Patients often receive 100% oxygen before transfer, and it is not uncommon for the oxygen flowmeter to be left on between inductions. Total time in the induction room usually approximates 5 min.
During an anesthetic induction being conducted in the operating room, a loud bang (described as sounding like a gunshot or a firecracker) was heard from the adjacent induction room. On opening the door to the induction room flames were observed within the carbon dioxide absorber canister. While the operating room staff cared for the now-anesthetized child, a nurse entered the induction room and turned off the oxygen flow (set at 1.5 l/min) and disconnected the gas supply lines from the wall source. A visitor to the facility turned off the sevoflurane (Ultane®, Abbott, North Chicago, IL) vaporizer (set at 8%) and a member of the surgical house staff attempted to put out the fire with a fire extinguisher. The still smoldering absorbent canisters were then placed under running water to completely extinguish and cool.
As soon as the brief surgical procedure was completed, the operating room was evacuated. The area was secured upon the arrival of firefighting personnel and remained unused until it had been decontaminated. The Ohmeda Excel 210 anesthesia machine and its two-canister GMS absorber assembly (Datex-Ohmeda, Madison, WI) had neither hoses nor bag attached at the time of the event. Examination of the machine revealed that the clear dome cover of the expiratory valve had fractured all around its rim and the central portion had been blown across the room (fig. 1). Significant charring and soot was noted on and within the absorber assembly. Melting and deformation of the top of the lower absorbent cartridge and the bottom of the upper absorbent cartridge had occurred, suggesting that the explosion and fire had originated in the upper part of the lower cartridge (fig. 2).
The absorber canister had contained two prefilled cartridges of Baralyme® (Allied Healthcare Products, St. Louis, MO), which had last been changed 11 days before the event. The induction room had been used four times the morning of the event; the oxygen flowmeter and sevoflurane vaporizer were known to have been turned off after the first two inductions (performed around 7:30 am and 8:15 am) but were certainly left on after the fourth induction (around 10:45 am) and possibly after the third (around 10:10 am), induction. As the fire occurred at 11:20 am, the total time that the absorber was exposed to oxygen and sevoflurane was between 35 and 70 min.
On the basis of the suspicion that the fire was related to a reaction between sevoflurane and desiccated barium hydroxide lime, anesthesia technicians were directed to change the absorbent in all the anesthesia machines in the facility. During this exchange, a second induction room anesthesia machine was discovered with melting and deformation of plastic components in the absorber canister suggesting extreme heat (fig. 3).
We were extremely fortunate that the anesthetic induction being conducted at the time of the event was performed in the operating room. Had the induction taken place in the induction room, the explosion and fire combined with a high flow of oxidant (oxygen and nitrous oxide) might have ignited the plastic components of the breathing circuit, resulting in significant injury to patient, parent, and staff.
Degradation reactions leading to the formation of compound A and related compounds from sevoflurane are reasonably well understood.1,6,8However, a report from Germany of five cases of prolonged or failed inhalation induction with sevoflurane suggested that degradation reactions involving sevoflurane might be considerably more complex than previously appreciated.4Because of the architecture of the circuits used in these cases, fresh gas was first directed through the absorber canister, rather than to the patient. The observation that end-tidal sevoflurane concentrations were much less than the vaporizer settings led to the suspicion that the vaporizer had malfunctioned; however, it later became clear that essentially all of the delivered sevoflurane was being degraded by desiccated absorbent. This degradation reaction also produced heat sufficient to make the absorber canisters hot to the touch and produce pungent and irritating gases causing agitation and coughing and rapid absorbent color changes, presumably from acidic by-products. Finally, two of the cases developed significant carboxyhemoglobin levels after the prolonged induction.
In the presence of strong base, sevoflurane can degrade to compound A, through elimination of hydrogen and fluoride to form a double bond, and to hexafluoroisopropanol and formaldehyde.1,6,8Formaldehyde can further react in the presence of strong base to form methanol and formic acid via the Cannizzaro disproportionation.6Methanol and formaldehyde are found among the degradation byproducts of sevoflurane.3,6Significant carbon monoxide (CO) production from the degradation of sevoflurane is seen as the reaction temperature exceeds 80°C.3,9It is speculated that there are at least two separate sets of degradation reactions involving sevoflurane: an initial, exothermic reaction which generates compound A, but not CO, and a second reaction which requires higher temperatures and produces CO.9
The degradation of sevoflurane is dependent upon the physical characteristics of the carbon dioxide absorbent to which it is exposed. Degradation increases as the water content of the CO2absorbent decreases, becoming maximal with complete desiccation.2,6,10,11Anesthetic degradation is also dependent on the strong, monovalent alkali content of the CO2absorbent (KOH and NaOH).6,8,10,12Complete desiccation requires several hours in a heated oven12or several days of exposure to a constant flow of dry gas at room temperature.4,5Newer absorbents that do not contain KOH or NaOH appear to greatly diminish or eliminate the degradation of anesthetics.1,10,12The absorptive capacity of absorbents without strong alkali is modestly decreased compared with that of conventional absorbents.10,12
Degradation reactions involving sevoflurane are considerably more energetic than those of the other halogenated ethers. Under laboratory conditions, the reaction of sevoflurane with desiccated absorbents containing strong monovalent alkali has routinely reached temperatures well in excess of 100–120°C,2–5,7,9while the reactions of enflurane, isoflurane, and desflurane have been associated with temperatures in the 50–80°C range.7,9Some experiments involving prolonged exposure of sevoflurane to completely desiccated absorbent have resulted in temperatures in excess of 350–400°C associated with smoldering, melting of plastic components, and even explosion and fire.2,3,5
The fire we experienced can therefore be explained by the degradation of sevoflurane in the presence of dry barium hydroxide lime, an absorbent with a high KOH content; this reaction resulted in the generation of extreme heat and volatile, flammable byproducts. Indeed, our event is eerily similar to the experiment reported by Holak et. al. , in which a two-canister absorber filled with desiccated barium hydroxide lime exploded after 53 min of exposure to 6 l/min of oxygen and 8% sevoflurane.3It is likely that our absorbent had become completely desiccated; it had not been changed in 11 days, had been used only briefly for induction of anesthesia using high gas flows, and had undoubtedly been exposed on multiple occasions to a continuous flow of fresh gas when operators failed to turn off the flowmeters after induction. In contrast to the case reports from Baum et. al. ,4the fresh gas in our circuit was delivered downstream from the absorber canister; fresh gas could either flow out through the inspiratory valve or retrograde through the absorber canister. Examination of absorbent drying with similar circuit architecture has shown that drying, and reactivity, is greatest at the exit from the canister.5Review of our institutional practice suggests several additional areas where absorbent desiccation may be a concern, including infrequently used machines in off-site locations, limited rehydration of absorbent resulting from the use of high fresh gas flows relative to body size, procedures (e.g. , myringotomy and tubes) having short durations that preclude rehydration of absorbent, and the use of high-flow insufflation techniques for endoscopy.
Given these concerns, it is perhaps surprising that we have not experienced additional instances of excessive heat generation or fire. Similarly, not all experiments involving degradation of sevoflurane by completely desiccated absorbents have resulted in fire. There are undoubtedly additional factors that help determine the risk of explosion and fire related to the degradation of sevoflurane. These factors may include a greater absolute mass of absorbent (catalyst),11a gas flow rate high enough to deliver a sufficient quantity of sevoflurane, but not so high as to wash reactants out of the canister, and preferential flow (channeling) resulting in localized desiccated absorbent and high reactant concentrations.
Our response to this event has been multifactorial. Anesthetists and technicians have been encouraged to turn off vaporizers and gas flows at the completion of each induction or anesthetic and instructed to change the absorbent if unattended gas delivery is encountered. CO2absorbent changes are now performed on a biweekly basis (more frequently in high-risk locations), rather than only as needed to avoid exhaustion of absorptive capacity. We initially substituted soda lime for barium hydroxide lime and are currently evaluating Amsorb® (Armstrong Medical, Coleraine, Northern Ireland), a CO2absorbent without strong alkali; if it meets our clinical requirements, Amsorb® will be introduced system-wide. Consideration was given to elimination of absorbent from the canisters in the induction rooms; although the clinical impact, given the use of high fresh gas flows and brief exposure to possible rebreathing, was deemed minimal, this intervention has been deferred for the time being.
In conclusion, we have presented an explosion and fire caused by the degradation of sevoflurane in the presence of desiccated CO2absorbent containing strong alkali. It is important to emphasize that the conditions that led to our event are identical to those that accentuate the generation of CO from the degradation of enflurane, isoflurane, and desflurane. Thus, even institutions that do not use sevoflurane must attend to the issues of absorbent content, desiccation, and anesthetic degradation. Efforts should be directed to avoiding desiccation of absorbents, combined with a low threshold of suspicion for changing absorbents and consideration of scheduled absorbent changes in high-risk locations. Newer absorbents free of strong, monovalent alkali greatly reduce degradation of anesthetics, and should be considered for high-risk locations or based on institutional practice. The byproducts of anesthetic degradation, whether CO, compound A, or heat and flammable volatiles, are insidious; avoidance of patient harm requires proactive elimination of factors which enhance anesthetic degradation rather than dependence on identification of degradation when it occurs.
The authors wish to thank Andreas Loepke, M.D., Assistant Professor, Clinical Anesthesia and Pediatrics, Department of Anesthesia, Children’s Hospital Medical Center, Cincinnati, Ohio, for his thoughtful translations from German to English.