Volume 9, No. 4 • Winter 1994

Safety Topics Prominently Featured at ASA

Susan K. Palmer, M.D.; Douglas C. Guyton, M.D.; Robert C. Morell, M.D.; Gerald L. Zeitlin, M.D.

At the annual meeting of the American Society of Anesthesiologists in San Francisco October 15-19, patient safety was again the underlying theme of a significant number of scientific presentations.

For example, a scientific poster session was held Wednesday morning, October 19, during which 29 abstracts were presented regarding patient safety, epidemiology, history and education. Topics included anesthesia simulators, postoperative nausea and vomiting, the safety of sevoflurane, and several outcome indicators related to coronary artery bypass surgery.

Anesthesia Simulators

Three posters dealt with the use of anesthesia simulators. Members of the Department of Anesthesiology from the University of Copenhagen at Herlev, Denmark, led by Dr. A. Lindekaer, studied the performance of eight anesthesiologists in the simulated scenario of unexpected ventricular fibrillation. Although the principles of ACLS were largely followed, the subjects were inconsistent in their performance and case management.

Dr. D. Gaba and colleagues, from Stanford and the Palo Alto VA Medical Center, examined interrater reliability of performance assessment during simulated episodes of cardiac arrest and malignant hyperthermia. Their results indicate that a single rater had a much greater chance of differing significantly from the mean score of five raters than did the score of a team of two raters. They suggested that evaluation reliability might be improved if:

1. raters are paired; 2. discussion is allowed prior to rating performance; 3. a simplified rating scale is used; and 4. specific performance behaviors are defined.

A survey of anesthesiologists’ attitudes toward simulators was conducted by Dr. R. Riley and colleagues from Royal Perth Hospital in Australia and the University of Pittsburgh. This study surveyed 181 Australian and American anesthesiologists. Results revealed that 73% of staff were in favor of departmental purchase of a simulator and 69% were willing to undergo evaluation using a simulator. Only 28% were supportive of compulsory simulator testing for re-certification. Many were concerned with cost.

Antiemetic Progress

Dr. T. Grasela and colleagues (part of a multicenter phase IV study and affiliated with SUNY in Buffalo, NY) reported an observational study on the effect of anticholinesterase agents on the incidence of postoperative nausea and vomiting (PONV). This study was conducted at 391 hospitals and included 6,908 patients. The authors found a slight increase in PONV in patients who received either neostigmine or pyridostigmine compared to patients receiving no anticholinesterase. They also noted a significant decrease in PONV in those patients receiving edrophonium. Dr. R. Mirakhur et al. (The Queen’s University and the Royal Victoria Hospital in Belfast, Ireland, as well as the University of Oklahoma) presented data which demonstrated the inability of atropine or glycopyrrolate to decrease nausea and vomiting after strabismus surgery, despite their success in inhibiting the oculocardiac reflex. P. Diemunsch and colleagues conducted a multi-center dose ranging study of the ability of granisetron, a 5-HT3 antagonist, to prevent postoperative nausea and vomiting. This study was sponsored by SmithKline Beecham and involved 35 medical centers in Europe and South Africa. Compared to placebo, granisetron was found to be effective at post induction doses of I and 3 mg.

Renal Safety of Sevoflurane

T. Stickler and colleagues (Abbott Labs) described the incidence of inorganic fluoride concentrations > 50 mol/L (considered by many to be the threshold blood level for concern about nonoliguric renal failure) in children and adults who were anesthetized with sevoflurane or reference agents (halothane, isoflurane or enflurane). 1,771 patients were included in the study in which 1,174 patients received sevoflurane and 597 received reference agents. 58 adults (8.1%) of adult patients in the sevoflurane group had inorganic fluoride concentrations @ 50 mol/L. No children nor any patients in the reference groups had inorganic fluoride levels exceeding the 50 mol/L threshold. No cases of renal failure were observed.


J. Ramsay from Emory University, working with co-authors from UCSF and The Medical College of Wisconsin, examined the association between Q-wave myocardial infarction (MI) and CPK-MB elevation after elective coronary artery bypass surgery. They found a weak relationship between these two indicators. Depending on the definition of myocardial infarction and the threshold for CPK-MB elevation, the incidence of a diagnosis of postoperative MI could vary from > 20% to < 2%. They believe that this variability could be an important consideration in outcome studies. A. Aggarwai’s group (same affiliations) examined the levels of CPK-MB following coronary bypass surgery at 15 medical centers throughout the United States. They presented the incidence of peak serum CPK-MB exceeding different threshold values. Their results indicate that release of this enzyme occurs as a continuum rather than occurring as a threshold phenomenon. These authors believe that this data may also have importance in the interpretation of complication and outcome studies.

Dr. K. Tuman and colleagues from Rush University, UCSF, the University of New Mexico and Harvard University reported on the effect of preoperative 8-blocker and calcium channel blocker therapy on the incidence of myocardial infarction and death after coronary artery bypass surgery. 2,284 patients undergoing CABG procedures at 24 medical centers were included in their database. The authors were not able to identify a beneficial outcome effect of either of these preoperative therapies on the incidence of postoperative MI or death. They speculate that these preoperative medications may not have adequate protective effects during and following the period of cardiopulmonary bypass. This study did not examine the effect of intraoperative or postoperative administration of either 8-blockers or calcium channel blockers.

Cardiac Anesthesia and Monitoring

Researchers at the Cleveland Clinic were able to use a unique cardiac data registry to examine the relationship between post-cardiac surgery arrhythmias and mortality. The registry of information was compiled concurrently during each patient’s intensive care unit stay. Data concerning serious arrhythmias in more than 3,000 patients showed that postoperative occurrence of a serious arrhythmia was a marker for higher mortality (15.6%) versus those patients who had no serious arrhythmias (1.7%). Female patients showed a greater incidence of arrhythmias (22.6%) compared to male patients (16.4%) even though other markers of cardiac disease were comparable. Mortality increased in the following order: supraventricular tachycardia, 13.5% mortality; heart block, 16.2%; ventricular tachycardia, 25.5%; and asystole, 74%.

Pediatric anesthesiologists at New England Medical Center in Boston studied the effect of transesophageal echocardiography (TEE) on hemodynamics in small infants undergoing cardiac procedures. They were able to document the disappearance of an arterial waveform in an extremity or the fall of blood pressure or oxygen saturation after placement of the TEE probe in infants weighing less than seven kilograms with cyanotic heart malformations. They recommend monitoring oxygen saturation in all extremities and consideration of removal of TEE probes during cardiopulmonary bypass if oxygen saturation or blood pressure falls in any extremity.

Anesthesiologists at Hahnemann University found that deflation of a thigh tourniquet routinely produces a shower of venous embolism greater than 3 mm in diameter detectable by TEE. Their study of embolic phenomena during two surgical techniques for total knee replacement showed that right ventricular embolism were almost always detected after tourniquet deflation. The period immediately following tourniquet deflation may present special risks for some patients since emboli were also seen in the left ventricle of 3 out of 56 patients studied.

Regional Blocks

Quality improvement data gathered during all anesthetic cases at the University of Colorado Health Sciences Center were examined to gain information about common problems with spinal and epidural blocks. The most common complication of 2,368 regional blocks performed in a 12-month period was failure of the blocks to anesthetize adequately for the planned surgery, which occurred in 4.8% of spinals and 3.8% of epidurals. Failure of epidural anesthesia occurred slightly more often in the older and sicker general operating room patients (7.7%) than in the obstetric patients (2.6%). Concerning complications, there was only one toxic reaction, two high blocks, and two extended duration blocks. No long term problems were recorded. Knowledge of realistic potential failure rates for major regional blocks can help practitioners and patients plan and prepare for alternate anesthetic methods.

Meta-analysis (statistically combining several similar studies) has become popular because it may demonstrate relationships or significant differences among clinical techniques which individual studies could not confirm. Anesthesiologists at the University of Cincinnati College of Medicine analyzed qualifying studies comparing pencil-point spinal needles with Quinke spinal needles to show that the Quinke-type needles were three times more likely to cause post-dural puncture headaches.

Problems with Intubations and Resuscitation

Difficult intubations are not easy to predict with precision, but patients who present for uvuloplasty are at least 24 times more likely to be impossible to intubate than the general population of operating room patients, according to Dr. B. Shapiro and colleagues at Northwestern University Medical School. The uvuloplasty patients studied were not morbidly obese but were still impossible to intubate in 3.8% of cases and difficult to intubate in 13.5% of cases.

Videotape recordings of intubations and resuscitation were used to analyze errors in technique by Dr. N. Jefferies and colleagues at the University of Maryland. Video was able to record improper cricoid pressure or neck stabilization techniques as well as difficulties with ventilator set-ups. When asked to self report errors during these resuscitations, practitioners tended to report adverse outcomes rather than procedural (process) errors.

At Massachusetts General Hospital, a study of emergency resuscitation found that managing the families of arrested patients and identifying the leader of the code teams were frequent problems. Missing supplies and/or malfunctioning equipment also plagued at least one third of the resuscitations. Although the identified problems were not contributory, a 30% immediate death rate following attempted resuscitation was recorded for these inhospital arrests.

At Dalhousie University, a study was conducted of the efficiency of airway management during in-hospital arrests. The plastic disposable laryngoscopes stocked in code carts were discovered to provide less-than-adequate visualization and even fractured on occasion. The study also found that small endotracheal tubes (6.0 mm or less) were needed and seldom available.

Using a bench set-up to simulate patient tissue resistances, Dr. S. Glowacki of Jefferson Medical College tried to induce current flow in a pulmonary artery (PA) catheter using a standard electrocautery unit. Although measurable current could be induced between the electrodes of the ejection fraction PA catheter, it was not clear that sufficient current could be induced in this manner to be a cause of arrhythmias with the use of these two devices together in the usual operating room setting.


The importance of education and vigilance in patient safety was underscored by a study performed by Drs. R. Leyssius and C. Kalkman from The Netherlands. In the first phase of their study, they determined the incidence, severity and duration of hypoxemic episodes in 450 PACU patients, with the PACU personnel unaware of the data collection. In phase two, they repeated their study with the PACU personnel aware of the data collection, and with an explicit written instruction to prevent desaturation below 90%. Reductions in incidence and severity of desaturation of 25-40% from Phase I to Phase H were reported. Four to six months after the study, data was again collected without the knowledge of PACU personnel (Phase HI). Unfortunately, the results of Phase III showed little difference from Phase 1, indicating that the improvements caused by increased awareness and vigilance during Phase 11 were not long lasting. It was felt to be of interest to see if the improvements occurring in Phase 11 could be sustained during a program of continuous education and increased awareness.

Computerization figured prominently in the 1994 meeting. Frequently cited reasons for reliance on computers included reduced costs (and even increased remuneration!), increased speed of information dissemination, and improved manipulation of data. “The automated record” stood as the goal in many presenters’ not-too-distant future.

Reporting from the University of Florida Dr. G. Gibby demonstrated improved capture of comorbidity diagnoses using a computerized pre-op evaluation form. Review of the computerized pre-op form resulted in reporting additional diagnoses in 22/180 records, and altered DRG coding in 5/180. Improved DRG coding results in more accurate epidemiologic data used for reporting purposes and scientific investigation. Dr. Gibby related the improved capture of comorbidity diagnoses not only to the readable, retrievable form that doesn’t forget or omit, but also to the anesthesiologist’s ‘fresh perspective caring for the total patient’ that results in a more complete listing of both acute and chronic medical problems. Another benefit was an additional $41,492 (a 1.5% increase) in reimbursement as a result of more accurate coding!

Dr. L. Mark, Johns Hopkins University, reported the development of an “Anesthesiology Consultant Report,’ designed to facilitate communication between the anesthesiologist and primary physician. An alternative to patients as the messengers (“Something happened during my surgery, somebody said my heart stopped.’), the report establishes a direct physician-to-physician link, legibly summarizing specific anesthetic concerns and perioperative events that may be of importance to the patient’s future health care. The negative aspects of an additional form ‘to unravel the mysteries of the anesthetic record’ were offset by the high rate of acceptance (82%) among the surgeons and primary we physicians to whom the report was delivered. Personal computers increased the speed and reduced the cost of producing these reports.

Dr. D. Doblar from the University of Alabama at Birmingham developed a scanner-based Continuous Quality Improvement System utilizing a personal computer. Such a system provides the flexibility to create practice-specific CQI indicators with ease. Rapid access to continuously updated CQI data risk adjusted to the specific patient population makes current individual and department profiles immediately available. Other advantages include reduced labor costs for data collection and processing.

Dr. R. Westerlund, New York University, validated the quality assurance process with a report on 65,008 cases from his institution. Consolidating input from QA forms, patient and physician reports, anesthetic records, as well as from ‘word of mouth,’ Dr. Westerlund utilized the QA process to identify and analyze problems. The QA process focused education in specific areas and appeared to lead to decreases in PACU reintubations, unplanned extubations, pulmonary edema, PACU respiratory depression, and allergic reactions to latex. Cooperative efforts with the pharmacy department led to decreased likelihood of drug misidentification due to look-alike ampules, and joint ventures with the surgery department resulted in better fluid management and blood product utilization.

Examining the reliability of QA data acquisition, investigators at the Massachusetts General Hospital looked specifically at inpatient units’ reporting of adverse drug events (ADE) that resulted in patient injury. They determined that reliance on Incident Reports alone was fraught with error, fading to discover 94% of cases of patient injury. Pilot studies showed voluntary submission captured 10% of events, while intensive chart review captured 90%. They suggested computerized ordering to minimize illegibility and transcription errors, facilitation of anonymous reporting of events, and ‘self-correcting teams” that find their own solution to problems and report to administrators.

Airway Alert

Dr. L. Mark of Johns Hopkins reviewed the first III patients enrolled in the Medic Alert Difficult Airway/intubation Registry. Verifying the Registry’s utility, she showed that anticipated difficulty with airway management was associated with significantly fewer intubation techniques (2.3/patient) and fewer adverse events (22%) when compared to unanticipated difficulty (3.7 techniques/patient; 44% adverse events). Interestingly, conventional laryngoscopy was successful in less than 20% of patients with difficult airways, prompting Dr. Mark to suggest that when difficult laryngoscopy/difficult airway is encountered, ‘changing technique may be more beneficial than changing blades.’ More reports will be forthcoming, as more patients are added to the database and as registered patients reenter the healthcare system for additional treatment.

Follow-up surveys of 61 patients enrolled in the Medic Alert Difficult Airway/Intubation Registry showed 100% satisfaction with the overall process of enrollment, despite a high rate of adverse events (19/61, or 31%) in ” group. Each enrollment form includes text specifically designed to address patient concerns and answer the most frequently asked questions. In addition, enrollees are provided with a visible Medic-Alert emblem, a wallet card, and a 24-hour emergency response system including FAX transmission of airway management records. Physicians are urged by the investigators to refer patients to the registry as an important step in relaying vital information to family and health care providers, information that may prevent future disasters.

Dr. D. Mangano from University of California at San Francisco presented preliminary data from the McSPI Research Group demonstrating an association between adverse postoperative outcomes and increased hospital costs. Vascular complications and stroke lengthened hospital stay by 60-70%, and stroke, CHF, and global encephalopathy increased ICU utilization by 40-50%. Stated, but not addressed, was reduction of the incidence of adverse outcomes (and healthcare costs) by improving quality of care. This awaits further study and analysis.

Another study from the McSPI Research Group Institutions examined the utilization of blood products in coronary artery bypass surgery. In a group of patients selected to be at low risk for transfusion there was good agreement among the 23 institutions for transfusion when the hemoglobin level was low; however, marked disparity in use of blood products was evident as the higher hemoglobin values were encountered. The incidences of complications and outcome data were lacking. However, marked differences in risk of complications from blood transfusion as well as inconsistent allocation of limited blood bank resources have obvious patient safety implications. The authors concluded that, despite the issuance of national consensus statements between 1985-88, substantial variability of practice persists, and a more aggressive approach to rational transfusion practice is warranted.

In a retrospective analysis, investigators at the Mayo Clinic estimated the incidence of dental injuries that led to dental consultation and recommendation for intervention/repair. Such injury occurred in 1/3,064 general anesthetics and 1/41,938 local or regional anesthetics with intravenous sedation. In each group injury was most commonly associated with intubation and use of oral airways, respectively, and the cost of repair ranged from $18 to $3,603. Risk factors for injury included general anesthesia with tracheal intubation, preexisting poor dentition, and increased difficulty of intubation. Limited cervical motion was a prominent feature of the patient with a difficult airway. Interestingly, the level of training of anesthesia personnel did not correlate with the incidence of tooth injury.

There were several posters of considerable interest that in past years would not have been thought of as being core patient safety issues; they illustrate the extraordinarily rapid shifts in ‘spearhead’ issues affecting the practice of anesthesia.

Four posters dealt with various aspects of sterility and prevention of infection. Dr. J. Scharf and colleagues from the University of South Florida examined again the question of whether perioperative blood transfusion, by suppressing the patient’s immune function, produced increased rates of post surgical infection. Despite using antibiotics routinely and microaggregate filters in the blood administration equipment, they found significantly more deep wound infections, wound dehiscence and multiple infections in transfused than in nontransfused patients. They point out that there were many confounding factors (e.g. length of operation) but by using logistic regression they found only ASA Classification and transfusion to be reliable predictive factors for infection.

In a thoughtful laboratory study, Dr. P. Langevin and associates placed an external inoculum of staphylococcus Aureus in a Monoject syringe and observed the effects of multiple subsequent injections. One cc of a precisely prepared culture was placed into the barrel of a syringe above the plunger. With each depression of the plunger, they found rapidly increasing numbers of ‘colony forming units’ in the sterile liquid previously placed below the plunger seal. They therefore recommend that indeed a syringe of this type should not be used more than once in a case.

Dr. M. Brown and colleagues of Wayne State University asked whether the re-use of syringes with a T-port extension (with a back check valve) plugged into the main conduit of a blood-containing IV set is safe. An expired unit of blood was placed in a pressure infuser and maintained at 300 mm Hg. Six 10 n@ syringes containing saline were connected at various points along the IV line each via a T-port extension. Samples of this saline were taken at various times between 15 n-minutes and three hours. There was no blood contamination in any of the samples taken from the proximal portions of the T-port or the syringes themselves. This study has significant implications for both patient safety and cost savings.

Bacterial Breakthrough

Dr. A. Layon and associates found an interesting difference in the ability of bacterial species to survive contact with the soda lime used in C02 absorbers. They found that ‘ was killed after the briefest contact with soda lime granules whereas Mycobacterium Avium survived up to one hour in direct contact. Of equal interest is their additional finding that both bacterial species, in an aerosolized medium, passed readily through the soda lime, i.e., between the granules.

The authors of all the above four studies state that further study is required to clarify the implications for clinical practice.

Dr. Palmer is Professor of Anesthesiology at the University of Colorado Health Sciences Center, Denver; Dr. Guyton is Assistant Professor of Anesthesiology at the University of Mississippi SOM/Medical Center, Jackson; Dr. Morell is Assistant Professor of Anesthesiology at The Bowman Gray School of Medicine of Wake Forest University, Winston-Salem, NC, and Dr. Zeitlin is from the Brigham and Women’s Hospital, Boston, and is an Instructor in Anesthesia, Harvard Medical School.

Carbon Monoxide: What Should We Do?

Editor’s Note: Reports in the last two APSF Newsletters concerning CO toxicity from the interaction of halogenated ether anesthetics and C02 absorbents have received a great deal of attention. Note that the experimental conditions in which this phenomenon can be induced on a lab bench involve absorbents deliberately dried to a very abnormally low (or absent) water content a highly unusual situation. Questions have been raised by Newsletter readers as to what all this means, particularly for daffy anesthetic practice in the O.R. An experienced observer of anesthesiology who also has a background in related research provides perspective and context through his opinions on these developments:

by Ralph A. Epstein, M.D.

Clinical decisions are most difficult when there is incomplete scientific knowledge. This is currently the situation the clinician faces in deciding how to react, if at all, to recent reports of the potential threat of anesthesia-induced carbon monoxide toxicity.

Carbon monoxide is toxic in very low concentrations. The severity of toxicity depends both on the duration of exposure and on the concentration inhaled. Other influencing factors include the level of exercise and coexisting medical conditions. From a practical standpoint, exposure to greater than 50 ppm for eight hours should be avoided. Even brief exposure to greater that 200 ppm is considered hazardous. Carbon monoxide reacts with hemoglobin to form deoxyhemoglobin. Textbooks of environmental medicine provide standard tables which predict carboxyhemoglobin level as a function of carbon monoxide concentration in inspired air and of duration of exposure. (1) For example, exposure to 1,000 ppm for one hour would be expected to produce 30% carboxyhemoglobin.

The toxicity of carbon monoxide is by no means limited to direct replacement of oxygen on hemoglobin. Even a small concentration of carboxyhemoglobin causes the oxygen-hemoglobin dissociation curve to shift sharply to the left, thereby preventing the normal unloading of oxygen in the capillary bed. Thus, the patient can suffer tissue hypoxia while the blood partial pressure of oxygen is high. Moreover, carbon monoxide also poisons the cytochrome enzyme system, further depressing the utilization of oxygen.

It is difficult to know exactly at what carboxyhemoglobin levels mortality occurs because most victims receive therapeutic oxygen between the time of exposure and arrival at a health care facility where the carboxyhemoglobin level is determined. However, the peak level can be estimated from the known half W of carboxyhemoglobin. It is generally thought that death may result from carboxyhemoglobin levels of 50 percent in young healthy victims. Patients with underlying cardiovascular disease may be at risk from significantly lower levels.

In recent years, anesthesiologists have not concerned themselves very much with iatrogenic carbon monoxide poisoning. This is in part because closed circle anesthesia fell out of favor after cyclopropane disappeared from clinical use. With closed circle anesthesia, there had been occasional reports of carbon monoxide build up in the anesthetic circuit, especially in patients who had required significant blood transfusions. Smoking by the blood donor and hemoglobin destruction were thought to be factors.

Confusion and Concern

Reports and discussion of iatrogenic carbon monoxide toxicity in the last two issues of the APSF Newsletter have therefore caused both confusion and concern in the anesthesia community. This has been compounded by the fact that the Newsletter is not a peer-reviewed journal. The information was therefore presented as “news,’ rather than in sufficient detail to allow full critical review. Nevertheless, the practicing anesthesiologist needs to decide whether the facts, as they are now known, require changes in clinical practice.

In the Summer issue of the Newsletter, Lentz’ reported a case of unanticipated carbon monoxide toxicity. Although it was not stated until an explanatory note appeared in the Fall issue, the patient was receiving desflurane. The author astutely recognized that the case occurred Monday morning and postulated that something happened to the anesthesia system during prolonged disuse which ultimately led to the release of carbon monoxide. Without understanding the specific mechanism, he made the common sense (but probably incorrect) recommendation that, after a weekend of disuse, the anesthesia system should be purged with a high flow of fresh gas prior to use. This seemed reasonable, particularly because it did not appear that there would be any disadvantage from such a strategy.

In the same Summer issue, Moon’ reviewed 29 recent cases of high levels of carboxyhemoglobin from three institutions. AU occurred after the anesthesia machine had been unused for at least two days. Moon reported, however, that he was unable to detect clinically significant amounts of carbon monoxide after experimentally exposing anesthetic agents to soda lime. He postulated that carbon monoxide might be adsorbed by soda lime and then released after a period of disuse. Again, flushing the system with a high fresh gas flow seemed reasonable.

Other Monday Events

This was not the first time that an adverse event had been associated with an anesthesia machine that had been unused over a weekend. Over a decade ago, we had noted that inhalation inductions in children were sometimes prolonged on Monday mornings. (4) In a series of studies, we showed that, with our particular anesthetic circuit, the soda lime dried out if the fresh gases were inadvertently left flowing over the weekend. The dry soda lime then adsorbed huge quantities of inhalation agents, markedly slowing induction. The amount adsorbed was roughly proportional to the degree of absorbent drying.

In the Fall issue of the Newsletter, Fang and Eger (5)’ shed additional light on the carbon monoxide problem. In brief, they reported that desflurane, enflurane and isoflurane all broke down in the presence of purposely dried carbon dioxide absorbents, releasing very high concentrations of carbon monoxide. They mentioned (without presenting specific data) that halothane and sevoflurane produced very little carbon monoxide.

In this laboratory situation, the concentration of carbon monoxide produced was alarmingly high. With the worst case combination of desflurane and completely dry barium hydroxide lime, the carbon monoxide concentration was almost 20,000 ppm. Desflurane produced a carbon monoxide concentration of about 15 times isoflurane and barium hydroxide lime about two times soda lime. However, even the best case combination of isoflurane and completely dry soda lime produced a concentration of 500 ppm, about 10 times more than is reasonable for chronic exposure. Fortunately, even moderate degrees of hydration of the absorbents greatly decreased the concentration of carbon monoxide produced. No carbon monoxide was produced in the presence of normally hydrated absorbents with any anesthetic agent.

[t is difficult to know what to do with all of this data. Though much of it illuminates important concepts, the data is really tangential to the clinical question. For example, under what clinical conditions (if any) do absorbents become sufficiently dry to cause halogenated ethers to break down to form carbon monoxide? Flow rate of dry gases, duration of flow and the design details of actual absorbers in clinical use certainly must be considered. What is the role of details of the circuit design? When we Discovered in 1982 the adsorption of anesthetics by soda lime that had become inadvertently dry, we were using a pediatric circuit specifically designed to humidify anesthetic gases. The dry fresh gases entered this circuit between the inspiratory valve and the absorber; the ‘pop-off’ valve was located between the absorber and the expiratory valve. Thus, unlike a modem circuit, all of the dry gas had to pass through the soda lime, even if the circuit was detached from the patient. This is not true with most modem circuits. Moreover, how rapidly does each of the currently used absorbents give up water and how rapidly are they rehydrated by the release of water from neutralization of carbon dioxide and by adsorbing exhaled water vapor? Finally, we cannot even be completely certain at this point whether the carbon monoxide levels found in these patients were due solely to the mechanism discovered by Fang and Eger.

It would be easy to decide what to do if there were sufficient, confirmed, peer reviewed data. At a minimum, I would like to know the the course of carbon monoxide concentration measured on both sides of the absorber during clinical conditions. I would particularly be interested in the time course of carbon monoxide production during rehydration of the dried absorbent.

Still, the clinician needs to decide what best to do in the face of an incomplete understanding of this potentially dangerous phenomenon. I believe ” is the time for a measured response rather than for a radical change in clinical practice. We need to remember that modern anesthetics, as currently used with absorbents, are actually very safe. I would not switch immediately to sevoflurane or halothane. The former is not yet available in the United States and the latter, for understandable reasons, has been largely replaced with more modern drugs. I would not even discontinue the use of desflurane, when indicated, in favor of isoflurane, even though, under very specialized laboratory conditions using completely dry absorbents, desflurane produces 15 times more carbon monoxide than isoflurane. These laboratory conditions may represent only very rare and brief clinical conditions. Likewise, the difference between soda lime and bariurn hydroxide lime is relatively small, and it is not yet known whether this difference exists under more typical clinical conditions. Furthermore, I would not, at this point, redesign the anesthesia machine to guard against dry absorbent.

My reticence to quickly make large changes in practice does not reflect my lack of concern about carbon monoxide toxicity. On the contrary, I believe that academic and industrial research is urgently needed to better delineate the problem. Without question, additional data might very well alter my reticence, but I can only base my actions on data that is currently available.

I would, however, modify my behavior to adhere more carefully to good clinical practice. It is clear to me that there are several good reasons not to allow carbon dioxide absorbents to dry out. I would first make sure that I understood the details of the anesthetic circuit I was using. In the very unlikely event that I were using a circle system in which the fresh gas inlet was located between the expiratory check valve and the absorber (so that all of the fresh gas was forced through the absorbent even when the Wye piece was open), I would completely change the absorbent at the start of each day and avoid the prolonged use of high fresh gas flows.

If I were using a conventional circle system, I would simply take care to turn off the fresh gas at the end of each case. I would instruct my technicians to follow up on this in the event I forgot. If I could not be sure that this would be done reliably, I would change the absorbent more frequently, perhaps every Monday morning. I believe this is all that is required by the facts as we now know them.

Dr. Epstein, Professor of Anesthesiology at the University of Connecticut, is a member of the Editorial Board of the APSF Newsletter.


1. Kurt TL. Chemical Asphyxiants. In: Environmental and Occupational Medicine, 2nd Ed, edited by WN Rom. Boston: Little, Brown and Co., 1992, p 541.

2. Lentz RE. CO poisoning during anesthesia poses puzzles. APSF Newsletter 9:14-15,1994.

3. Moon RE. Cause of CO poisoning, relation to halogenated agents still not dear. APSF Newsletter 9:14-16,1994.

4. Grodin WK. Epstein RA. Halothane adsorption complicating the use of soda lime to humidify anesthetic gases. Brit. J. Anaesth. 54:555-559,1982.

5. Fang ZX. Eger El. Source of toxic CO explained: -CHF2 anesthetic + dry absorbent. APSF Newsletter 9:25-30,1994.