Volume 7, No. 1 • Spring 1992

APSF Sponsored Simulator Research Aims at Reducing Errors

Howard A. Schwid, M.D.; Daniel O'Donnell, Ph.D.

Editor’s Note: APSF Research Grant recipients are requested to summarize their funded projects upon completion.

by Howard A. Schwid, M.D. and Daniel O’Donnell, Ph.D.

It is estimated that human error is responsible for at least 70% of anesthesia events that lead to adverse outcome. Retrospective analysis of management of critical incidents is limited by difficulty in analyzing the response to the critical incident due to poor record keeping during the crisis, missed information, inaccurate observations, and incorrect sense of time. Due to the low frequency of critical incidents in the operating room, prospective evaluation of the response of a large number of anesthesiologists to these events is not practical.

In an effort to model the clinical setting, we used an anesthesia simulator to create critical incidents in order to observe how anesthesiologists diagnose and treat these situations. This approach provides an opportunity to look for patterns of errors in management. The purpose of this study was to assess the ability of anesthesiologists to recognize diagnostic clues, make the diagnosis rapidly, to affect treatment, and to evaluate the patient’s response during simulated critical incidents.

A screen based, graphical simulator that operates on IBM AT or compatible or Macintosh computer was used for the evaluation. The program is called the Anesthesia Simulator Consultant(ASC) and is an expanded version of SIMCORD, the Anesthesia Simulator-Recorder. In these programs, a graphical interface displays the simulated patient and monitors and allows the ,anesthesiologist to examine the patient, administer drugs, control the airway, ventilate, and administer fluids using mouse-controlled input. Mathematical models of physiology and the pharmacologic effects of 70 drugs predict the simulated patient’s responses. A variety of patient scenarios and critical incidents can be reproduced with the simulator. The patient’s vital signs and all management decisions are automatically recorded for review following the case.

Thirty anesthesiologists (10 residents, 10 faculty anesthesiologists, and 10 anesthesiologists in private practice) were evaluated on their management of six simulated cases. The residents had a minimum of one year of anesthesia training. The faculty anesthesiologists had an average of 7.6 years experience and the private practice anesthesiologists had an average of 8.5 years experience. The six cases were 1) a healthy patient with a full stomach; 2) an elderly, dehydrated patient; 3) an esophageal intubation; 4) a patient who develops myocardial ischemia intraoperatively; 5) an anaphylactic reaction; and 6) a cardiac arrest. None of the subjects knew the type of problem they would confront at the start of each simulation. The subjects were encouraged to vocalize their thoughts during the case; these comments were recorded manually and the individual’s actions using the simulator were recorded by the program.

The first two cases on the list did not include preprogrammed critical incidents and provided the subject the opportunity to become familiar with use of the simulator. The authors were present throughout all the simulations to assist subjects with operation of the program.

Many management errors were observed in the study of simulated critical incidents. Both experienced and inexperienced anesthesiologists made significant management errors.

The esophageal intubation was recognized by most subjects either by lack of carbon dioxide on the capnogram or by distant breath sounds with carbon dioxide analysis confirmation. Two residents missed the diagnosis. One thought it was severe bronchospasm and the other thought the capnometer malfunctioned. Myocardial ischemia was often inadequately treated. Tachycardia and hypotension were recognized but were frequently left untreated. In a few instances inappropriate drugs were administered including labetalol and sodium thiopental to treat tachycardia despite hypotension. Continuous infusions of vasoactive agents were especially troublesome. Typical therapeutic dose ranges were not known for a selected agent by 30% of subjects and 50% had difficulty calculating the correct infusion rate in drops/min-1 for the desired mcg/kg-1/min-1.

Anaphylaxis Missed

In the case of simulated anaphylactic reaction, the diagnosis of anaphylaxis was missed 60% of the time. The most frequent incorrect or incomplete diagnoses were supraventricular tachycardia, pneumothorax, and electromechanical dissociation. Several subjects (30%) including experienced clinicians undertreated severe hypotension while trying to find the etiology. In addition, many subjects assumed the tachycardia was due to inadequate anesthesia and administered additional anesthetic agents without checking the blood pressure. Frequent reassessment was found to reduce management errors since anesthesiologists who reevaluated blood pressure at three-minute intervals made significantly more therapeutic errors than those who measured blood pressure more frequently.

Most subjects were not able to recall the initial steps of the Advanced Cardiac Life Support algorithm for the treatment of ventricular fibrillation. The most common errors involved incorrect dosages of medications, incorrect defibrillator use, failure to hyperventilate, and failure to turn off the vaporizer. A clear relationship was found between management of the cardiac arrest and the time since the last ACLS training. Seventy-one per cent of anesthesiologists trained within six months of evaluation managed the arrest according to ACLS guidelines. This number decreased to 33%, 28%,, and 25% for those anesthesiologists with ACLS training in the prior 712 months, 13-18 months, and 19-24 months respectively. No anesthesiologist who had ACLS training more than 24 months prior to evaluation or who never had ACLS training was able to follow the first few steps of the ACLS protocol management of ventricular fibrillation.

Simulation vs. Life

It is recognized that performance in a simulator is different from performance in the real situation. In some cases, performance in a simulator will be better in the simulator and in some cases it will be better in the real world. Heightened vigilance may improve performance in a simulator while unrealistic simulation of the workplace may have a negative impact. Even in a full-scale simulator, fidelity of the mannequin has limitations. The real operating room environment is admittedly very different from the screen-based simulator. It is possible that anesthesiologists’ responses to real critical incidents in a real operating room could be different from this screen-based simulator, but we feel the observations concerning management of simulated critical incidents are valid despite the differences between the screen-based simulator and the real situation. Graphical simulation does not account for misinterpretation of absent carbon dioxide in the capnogram after intubation, treating tachycardia without determining the blood pressure, difficulty with calculating infusion rates, or inability to recall ACLS protocols. There is no reason to believe these errors would not have occurred in the operating room under similar circumstances.

Every subject evaluated in this study made at least one potentially dangerous error. Methods must be developed to improve anesthesiologists’ readiness for critical incident management. It may be helpful to have simple charts on the anesthesia machine for differential diagnosis and tables for continuous infusions of medications. Based on the poor retention of ACLS protocols six months after training, we propose that anesthesiologists review the management of critical incidents at least every six months. In addition to cardiac arrest, anesthesiologists should be prepared for anaphylaxis, difficulty in ventilating an intubated patient, malignant hyperthermia, pneumothorax, pulmonary embolism, equipment failures, and other emergencies. The preparation can consist of reading and reviewing an organized approach (algorithm) or taking refresher courses covering these incidents. Anesthesia simulators may become an important way to train and retrain for critical incidents. Anesthesiologists can now rehearse the management of these problems with a graphical simulator on a personal computer; in the next few years full-scale operating room simulators may become accessible.

Drs. Schwid, Associate Professor, and O’Donnell, Systems Analyst Programmer, are from the Department of Anesthesiology, University of Washington (V.A. Medical Center), Seattle. They were recipients of a 1990 APSF Research Grant.