Hazard [Health Devices Nov 1998;27(11):402-4]

ECRI has investigated several incidents of patient exposure to carbon monoxide (CO) during the administration of inhalation anesthetics through semiclosed circle anesthesia systems. In one of the incidents, patient injury resulted. In each case, after ruling out other possible sources of CO, we concluded that dangerous levels of the gas were generated within the anesthesia system under the conditions present during the incidents. These conditions included the presence of excessively dry carbon dioxide (CO₂) absorbent in an anesthesia system being used to deliver halogenated anesthetic agents for the first case of the day.

Similar incidents have been reported in the literature, with one common characteristic being the timing of the exposures. Many incidents have occurred during Monday morning cases, and all appear to be associated with the first delivery of an anesthetic after a lengthy period (e.g., overnight, over a weekend) of anesthesia machine inactivity.

Carbon monoxide is very toxic, even in low concentrations. Once in the blood, CO binds tightly with hemoglobin, forming carboxyhemoglobin and diminishing the ability of hemoglobin to transport and release oxygen. The level of CO exposure will be a function of both the inhaled concentration and the exposure duration. The specific effect on the patient will vary depending on the patient's cardiovascular condition and the level of oxygen administered before and during administration of the anesthetic.

To understand how CO exposures can occur, readers will need a basic understanding of circle anesthesia systems and the role of CO₂ absorbers within these systems. Inhalation anesthetics are usually administered through semiclosed circle anesthesia systems, although closed circle systems are sometimes used. In either type of circle anesthesia system, some portion of the gas exhaled by the patient is recirculated through the system and back to the patient, thus conserving medical gases, vaporous anesthetics, and expired water vapor.

To prevent dangerous levels of CO₂ from accumulating in the recirculating gas mixture, anesthesia machines that employ circle systems include an integral CO₂ absorber to remove the CO₂ exhaled by the patient. These absorbers typically consist of two stacked canisters containing granular absorbent materials that chemically neutralize CO₂ as the exhaled gas passes through. Commonly used absorbent materials include soda lime (e.g., Sodasorb) and barium hydroxide lime (e.g., Baralyme). When the ability of these materials to neutralize CO₂ becomes exhausted, the absorbent is replaced. For most absorbents, the current basis for determining when replacement is needed is the change in color of a pH indicator impregnated in the absorbent material.

Discussion

Although the exact chemical mechanism by which CO can be generated is not clear, published studies have indicated that a reaction between halogenated anesthetic agents and commonly used CO₂ absorbents can produce CO if the CO₂ absorbent is excessively dry. Drying out of the absorbent material can occur when 1) an anesthesia machine has been sitting idle, such as over a weekend, and 2) there is a continuous flow of medical gas (which is very dry) through the CO₂ absorber. When dry, the absorbent becomes highly reactive in the presence of certain halogenated agents, resulting in the production of CO as the agent flows through the machine's CO₂ absorber. Desflurane (Suprane) appears to be the most reactive of the halogenated anesthetic agents, although other agents—particularly enflurane and isoflurane—have also been reported to produce CO. The reaction between the agent and the absorbent material can continue for many minutes.

Complicating matters is the fact that identifying patient exposure to CO when it does occur can be difficult because carboxyhemoglobin levels are not monitored during anesthesia. Monitoring devices such as pulse oximeters and blood gas analyzers are not intended to detect carboxyhemoglobin; in fact, pulse oximeters will usually detect carboxyhemoglobin as oxyhemoglobin. Similarly, medical mass spectrometers are not configured to detect CO. And while whole blood co-oximeters can distinguish carboxyhemoglobin from oxyhemoglobin, these devices require a fresh blood sample and cannot provide real-time monitoring. As a result, CO exposure may go undiscovered unless patient morbidity leads to a comprehensive clinical and device investigation.

In the cases investigated by ECRI, anesthetists identified all the incidents of CO exposure indirectly. For example, in the incident that resulted in an injury, the patient's pulse oximetry readings had become erratic, but the heart rate and ECG waveform remained normal. After the same results were obtained using another pulse oximeter (of the same model) and a new probe, blood was drawn for a blood gas analysis, which revealed a high partial pressure of oxygen (PaO₂ greater than 600 mm Hg). Suspecting a problem with the anesthesia machine, the staff switched to a different machine. The blood sample was then analyzed by co-oximetry, which revealed a carboxyhemoglobin level of 60% to 70% (values that grossly exceed normal levels); thus, the cause of the patient's condition was determined to be CO exposure.

One further complication is that it can be difficult to determine when CO exposure is likely to occur because there appears to be no readily available, convenient, or reliable means of monitoring moisture within an absorber or of rehydrating absorbent that has dried out. Thus, to prevent the conditions under which CO can be produced from developing, users will need to ensure that the absorbent does not dry out. To do this, they need to ensure that the flow of medical gas is discontinued whenever an anesthesia machine is not in use on a patient; it is particularly important that the gas flow be stopped at the end of the workday.

It should be stressed that the reactions that produce CO within an anesthesia system do not occur while the machine is idle; rather, they occur only when agent vapor flows through the absorber. Therefore, flushing the breathing circuit with fresh gas before use (such as during a pre-use check) will not prevent or relieve the problem. It should also be stressed that CO exposures are unlikely to be detected intraoperatively; thus, healthcare facilities need to ensure that the conditions under which CO can be produced during inhalation anesthesia do not occur. Specifically, users must be sure to discontinue the flow of medical gas whenever an anesthesia machine will not be promptly used on another patient. ECRI recommends that the absorbent material in both canisters of an absorber be replaced whenever there is reason to believe that a machine has been left idle with gas flowing for an undetermined time. Fresh absorbent materials are sufficiently hydrated and normally remain hydrated by exhaled water vapor in the circle system, thereby preventing reaction with halogenated agents.

There is still much to be learned both chemically and clinically about the phenomenon of CO production associated with the interaction of halogenated anesthetic agents and CO₂ absorbent materials. ECRI will continue to assess relevant new findings in the medical literature and to evaluate changes in anesthesia monitoring and delivery systems. Given the present technology and knowledge of the problem, all efforts to prevent CO exposure must be directed at detecting and protecting against unintended medical gas flow when anesthesia systems are not in use.

1. Alert anesthesia and other appropriate personnel to the problem and to our report.
2. Ensure that medical gas is turned off when an anesthesia machine will not be promptly used for another procedure. At the end of every day, verify that the gas is off for all machines.
3. Before performing a pre-use check for the first case of a day, determine if there is any flow of medical gas. If there is, replace the absorbent material in both absorbent canisters before using the machine. Identify and address the cause of the gas flow.

Barker SJ, Tremper KK. The effect of carbon monoxide inhalation on pulse oximetry and transcutaneous PO₂. Anesthesiology 1987 May;66(5):677-9.

ECRI. Pre-use checklist for anesthesia units (machines and accessories). Health Devices 1988 Sep;17(9):275.

ECRI. Sodasorb PrePak CO₂ absorption cartridges [hazard]. Health Devices 1988 Jan;17(1):35-6.

Epstein RA. In my opinion: carbon monoxide: what should be done? J Clin Monit 1995 May;11(3):217-9.

Fang ZX, Eger EI 2nd, Laster MJ, et al. Carbon monoxide production from degradation of desflurane, enflurane, isoflurane, halothane, and sevoflurane by sodalime and baralyme. Anesth Analg 1995 Jun;80(6):1187-93.

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 Aug;87(2):308-16.

Woehlck HJ, Dunning M 3rd, Connolly LA. Reduction in the incidence of carbon monoxide exposures in humans undergoing general anesthesia. Anesthesiology 1997 Aug;87(2):228-34.

Woehlck HJ, Dunning M 3rd, Gandhi S, et al. Indirect detection of intraoperative carbon monoxide exposure by mass spectrometry during isoflurane anesthesia. Anesthesiology 1995 Jul;83(1): 214-7.

• Anesthesia Unit Absorbers, Carbon Dioxide [10-140]
• Anesthesia Unit Carbon Dioxide Absorbents [17-509]
• Anesthesia Units [10-134]

Device factor: Device interaction

User errors: Incorrect clinical use; Incorrect control settings

External factor: Medical gas and vacuum supplies

Exposure to hazardous gas; Suffocation