Canister Fires Become A Hot Safety Concern

Michael A. Olympio, MD; Robert C. Morell, MD

Reports of fire and/or extreme heat occurring in the carbon dioxide absorber portion of the anesthesia circle system have come to the attention of the APSF. An October communication received from an anesthesiologist described canister overheating and a burning expiratory valve. Rapid communications and discussions revealed the existence of other, extremely rare, but similar occurrences. Input from the ASA Committee on Equipment and Facilities and from the FDA Center for Devices and Radiologic Health revealed 3-4 other reports. While the exact etiology of these “canister fires” is not known, the mechanism appears to be related to chemical interactions between desiccated CO2 absorbent and potent inhaled anesthetic agents. The ECRI has also received reports of this dangerous phenomenon and has identified some common elements in 4 fires that were reported to them over the past few years. These common elements include the use of barium hydroxide containing CO2< absorbent, desiccation of the absorbent, and the use of sevoflurane.

Abbott Laboratories issued a “Dear Health Care Professional” letter on November 17, 2003, calling attention to these rare, isolated reports.1 In their letter a number of suggestions are described that might limit the risk of canister fires. These include 1. If you suspect that the CO2 absorber may be desiccated because it has not been used for an extended period of time, it should be replaced. 2. Shut off the anesthesia machine (and fresh gas flow) after any case, when an extended period of non-use is anticipated. 3. Turn off the vaporizers when not in use. 4. Verify the integrity of the packaging of new CO2 absorbents prior to use. 5. Periodically monitor the temperature of the CO2 absorbent canisters. 6. Monitor the correlation between the sevoflurane vaporizer setting and the inspired sevoflurane concentration. An unusually delayed rise or unexpected decline of inspired sevoflurane concentration compared to the vaporizer setting may be associated with excessive heating of the CO2 absorbent canister.

Abbott also pointed out that the color indicator of CO2 absorbents does not necessarily change as a result of desiccation. If excessive heat is detected the patient should be disconnected from the anesthesia circuit, fresh gas flow to the circuit should be shut off, and the CO2 absorbent should be replaced. The patient should also be monitored for carbon monoxide exposure and the potential for chemical thermal injury. Clinical findings associated with these events can include 1. Failed inhalation induction or inadequate anesthesia with sevoflurane. 2. Clinical signs of airway irritation. 3. Oxygen desaturation, increased airway pressure, and difficulty with ventilation. 4. Severe airway edema and erythema. 5. Elevated carboxyhemoglobin levels.

It was further discussed that the cases of canister fire or extreme heat generation were typically the first case of the day and involved the use of barium hydroxide containing absorbent; however, other case reports have involved desiccated soda lime, as well. Abbott Laboratories, in collaboration with the Food and Drug Administration, is actively investigating the mechanisms of these events and associated factors. Any readers experiencing similar problems are strongly encouraged to report such events to the FDA’s MedWatch program (phone: 1-800-FDA-1088, Fax: 1-800-FDA-0178, or via electronic reporting at the FDA MedWatch Website at or to Abbott Laboratories (phone: 1-800-633-9110).

We call the reader’s attention to a most dramatic (American) report by Elena J. Holak, Harvey J. Woehlck, and colleagues at the Medical College of Wisconsin, in which simulated anesthetic conditions were created in the laboratory.2 The authors dehydrated the barium hydroxide containing CO2 absorbent and exposed it to an attempted maintenance of 1 MAC sevoflurane (ET 2.1%) in the presence of 350 ml/min of carbon dioxide “production.” The greater the minute ventilation (and presumably greater exposure of the delivered sevoflurane to the absorbent), the greater was the fresh gas flow (FGF) required to achieve the 2.1% target. At a minute ventilation of 10 L, it required 6 lpm of FGF and a dial setting of 8% to achieve the target. In the absence of sevoflurane uptake by a patient, the high breakdown of sevoflurane was presumed secondary to reaction with the absorbent. In less than 10 minutes of exposure, the upper absorbent canister reached >110°C, and was too hot to touch in 15 minutes. CO production increased exponentially above 70°C, and at 45 minutes the temperature was >200°C, the upper limit of the thermometer. Finally, at 53 minutes, the absorber exploded and burst into flames. It was suggested that a delayed rate of rise of the inspired agent concentration could serve as an early warning before the dramatic rise in temperature of the absorbent. Temperature monitoring of the internal aspects of the absorbent (particularly the layer first exposed to agent) may represent a clinically useful tool to help detect the possibility of sevoflurane breakdown in the presence of desiccated absorbent.

Sevoflurane is flammable at a concentration of 11% in oxygen.3 However, the byproducts of sevoflurane breakdown include methanol and formaldehyde in addition to CO, and these may be potentially combustible.

The exposure to FGF and subsequent dehydration of absorbent within a particular anesthesia machine depends upon a complex set of interactions between flow, resistance, and unidirectional valves. One report describes the removal of the reservoir bag as a cause of increased retrograde flow and desiccation.4 However, modern anesthesia machines, with radically different circuits, may also have different paths of least resistance, and thus, fresh gas flow. Differences in APL (adjustable pressure limiting) valve design or entry of the FGF into the circuit distal to the inspiratory valve might also affect desiccation. Perhaps the most important precaution is to ensure that all FGF (including minimum flow) is off when the machine is not in use. Furthermore, it may not be desirable to completely turn off certain modern anesthesia machines because start-up delays could occur in emergency situations (by virtue of automated self-check procedures).5,6 Monitoring the frequency with which the CO2 absorbent is changed and developing site-specific policies for changing the absorbent may also be prudent. The safety of rehydration of absorbent7 has not been fully characterized.

Awareness and education are the first steps toward prevention. Numerous and important abstracts from the 2002 and 2003 meetings of the American Society of Anesthesiologists address this topic.8 The APSF will continue to communicate closely with clinicians, manufacturers, and other safety organizations to monitor developments and emerging information. It is our continuing goal to assist all anesthesia providers in their attempts to ensure patient safety by disseminating such important information via the APSF website ( and this Newsletter.

Dr. Olympio is Professor and Vice Chair for Education in the Department of Anesthesiology, Wake Forest University School of Medicine, Winston-Salem, NC. Dr. Morell, Editor of the APSF Newsletter, is in private practice in Niceville, FL, and Clinical Associate Professor of Anesthesiology at Wake Forest University School of Medicine.


1. Dear Health Care Professional (Letter from Abbott Laboratories, November 17, 2003). Available online at:

2. Holak EJ, Mei DA, Dunning MB III, et al. Carbon monoxide production from sevoflurane breakdown: modeling of exposures under clinical conditions. Anesth Analg 2003;96:757-64.

3. Wallin RF, Regan BM, Napoli MD, Stern IJ. Sevoflurane: a new inhalational anesthetic agent. Anesth Analg 1975;54:758-66.

4. Frink EJ, Nogami WM, Morgan SE, et al. High carboxyhemoglobin concentrations occur in swine during desflurane anesthesia in the presence of partially dried carbon dioxide absorbents. Anesthesiology 1997;87:308-16.

5. Gross JB. Draeger Narkomed 6000 poses patient safety risks (letter). Anesthesiology 2001;95:567.

6. Feldman JM. Draeger Narkomed 6000 poses patient safety risks (reply). Anesthesiology 2001;95:567-8.

7. Baxter PJ, Kharasch ED. Rehydration of desiccated baralyme prevents carbon monoxide formation from desflurane in an anesthesia machine. Anesthesiology 1997;86:1061-5.

8. American Society of Anesthesiologists. ASA Annual Meeting Abstracts Online. 2002 Abstracts: A-82, A-83, A-84, A-1081, A-1096, A-1155, A-1156, A-1181, A-1186. 2003 Abstracts: A-508, A-626, A-1238. Available on the web at: