British Journal of Anaesthesia 1997; 78: 637–641 CLINICAL INVESTIGATIONS Inaccurate reporting of simulated critical anaesthetic incidents A. J. BYRNE AND J. G. JONES Summary Eleven anaesthetists completed a simulated anaesthetic which was deliberately complicated by a slow progressive bradycardia followed by an episode of severe bronchospasm. After the simulation, each anaesthetist was asked to complete an anaesthetic chart and a critical incident report. Considerable discrepancies were found between the anaesthetists´ written accounts, a videotape of their performance and actual data from the simulator. During the simulations, all of the anaesthetists reacted appropriately and treated their “patient” successfully but their written accounts showed a tendency to record “typical” rather than actual events and to ignore events not consistent with their final diagnosis. Only four anaesthetists mentioned bradycardia in their written description and none accurately described the changes in arterial pressure during the episode of bronchospasm. The findings are in keeping with other studies which suggest that people record events as “schemata” rather than as collections of discrete facts. These results have significant implications for those involved in the teaching of anaesthesia and in the analysis of critical incidents. (Br. J. Anaesth. 1997; 78: 637–641). Key words Anaesthesia, audit. Computers, simulation. computer simulation. Anaesthetist, training. Model, The Anaesthetic Computer Controlled Emergency Situation Simulator (ACCESS)1 has been used over the past 3 yr to provide basic training for junior anaesthetists and to assess their ability to deal with critical incidents. The simulator consists of a resuscitation manikin in place of a patient, an anaesthetic machine/ventilator and computer simulated monitors. The anaesthetic machine is a basic “Boyles machine” complete with TEC3 vaporizer, which appears to function normally, but which has been modified to avoid the use of real volatile agents or nitrous oxide. A Manley MP3 ventilator was used during this study. The monitors provide information on non-invasive arterial pressure, pulse oximetry, ECG, inspired oxygen concentration and end-tidal carbon dioxide concentration. Simulations are presented as clinical scenarios which test the ability of the trainee to deal with a specific emergency. The simulations are realistic, as judged by those who have taken part,1 and require appropriate manual responses to be performed by the trainee, for example, hand ventilating the lungs of the “patient” or injecting drugs. During previous training sessions it was noticed that trainees often seamed to have misinterpreted the data presented and, when asked to describe their actions, their descriptions often contained factual inaccuracies. However, these inaccuracies were usually plausible and consistent with their stated final diagnosis. For example, if they failed to notice a gradual bradycardia, they often stated that heart rate had been normal initially and had then changed suddenly. This study was designed to test if the observed anomalies arose from the anaesthetist misunderstanding the simulation, from the anaesthetist’s inability to manage the simulated critical incident, or if the problem lay in the ability of the anaesthetist to remember events accurately. Methods Eleven trainee anaesthetists, with 3–8 yr clinical experience, volunteered to take part in a single simulation, with no opportunity for trial runs or practice. None had used ACCESS previously and they were told only that their performance would be evaluated during a standard simulated anaesthetic. The same investigator gave each anaesthetist a verbal introduction to the simulator lasting 3–4 min. During each simulation, the investigator acted as a helpful aide, to provide physical assistance for any task, but not to provide useful diagnostic or therapeutic information. However, the investigator did answer questions related to the simulation, for example, if cyanosis was present or if the breath sounds had altered in character. Each anaesthetist was introduced to the simulation as if taking over a routine anaesthetic from a colleague. They were shown a preoperative assessment form and were told about the anaesthetic technique being used. The “patient” for each anaesthetist was based on a real case and was young and healthy. The “operation” was an open reduction and internal fixation of an ankle fracture, the patient having had a local anaesthetic ankle block and been A. J. BYRNE, MRCP, FRCA, Department of Anaesthesia, Addenbrooke`s Hospital; J. GARETH JONES, MD, FRCP, FRCA, University of Cambridge and Addenbrooke`s Hospital; Hills Road, Cambridge CB2 2QQ. Accepted for publication: December 18, 1996. Correspondence to A. J. B. 638 paralysed, the trachea intubated and the lungs ventilated 20 min before the anaesthetist taking over the case. Drugs “used” were thiopentone, vecuronium, fentanyl and bupivacaine. Each anaesthetist was allowed to ask questions and to check the settings on the ventilator, anaesthetic machine and monitors. The simulation was started as soon as the anaesthetist was familiar with the situation and ended when the “patient” had been treated correctly and was stable. Each entire episode was videotaped for later analysis. At the end of each simulation, each anaesthetist was asked to immediately complete an anaesthetic chart and a critical incident report, covering the period of simulation. They were not asked questions about any specific event, but were simply asked to record the simulated events in free text, as if it were a real case. The computer printed out a timed list of all significant events and the heart rate of the “patient” during the period of each simulation. This printed output was not made available to the anaesthetist. Each simulation contained two critical incidents: phase one, a bradycardia and phase two, an episode of bronchospasm induced by a dose of antibiotic. During phase one, heart rate started at 80 beat min1 and decreased in a linear manner, by 10 beat min1 every 3 min, with only minor random variation. There was no consistent change in arterial pressure. It was expected that at some point the anaesthetist would administer a dose of anticholinergic drug to correct the bradycardia. When this dose had been given and bradycardia corrected, the simulation moved into phase two. The purpose of the first phase was to compare the actual change in heart rate (from the simulator printout) with the change documented on the anaesthetic chart and the anaesthetist’s perception of events from their written account. At the start of the second phase, the anaesthetist was told that the surgeon had requested that a dose of i.v. flucloxacillin be given. As soon as the anaesthetist administered some of the flucloxacillin, airways resistance increased markedly, simulating an episode of bronchospasm. This was manifest by the Manley ventilator being unable to deliver the set tidal volume. Arterial pressure remained normal for two further readings (at least 2 min) and then rapidly became unrecordable. This simulated a reduced cardiac output secondary to air trapping and was based on a real critical incident. It was expected that the anaesthetists would administer i.v. fluids and adrenaline, although any reasonable combinations of treatment would have resulted in the “recovery” of the “patient”. For example treatment with salbutamol, aminophylline or adrenaline, in appropriate doses, would all result in a reduction in airways resistance. When these treatments had been given and the “patient” had stabilized, it was planned that the simulation would be terminated. The purpose of this phase was to observe if the anaesthetists would be aware of the changes in arterial pressure, if these changes would influence diagnosis/treatment and if the information would be recorded in their written account. The realism of the simulations was assessed by British Journal of Anaesthesia asking the anaesthetists to answer two questions, using a linear analogue score from 1 (unrealistic) to 10 (realistic), in response to the questions “How do you rate the realism of the simulation?” and “How do you rate the realism of your responses to the simulation?”. This assessment was performed to ensure that ACCESS was providing a level of realism. Results All anaesthetists responded appropriately to the simulation; all “patients” were treated appropriately and all simulations were continued to a successful conclusion. RESPONSES TO SIMULATED BRADYCARDIA In the first phase, all 11 anaesthetists diagnosed bradycardia and administered an adequate dose of anticholinergic, either atropine 300–600 g or glycopyrronium 200–400 g. However, three anaesthetists failed to document bradycardia and two failed to record administration of a drug on their anaesthetic chart. In the critical incident report, only four of the 11 anaesthetists mentioned bradycardia, and these comments are reproduced in table 1. Only the first anaesthetist described the simulation accurately, with the next two anaesthetists being less specific. Anaesthetist No. 4 described the actual, gradual onset of bradycardia as a single rapid episode, with associated hypotension (which was not present). Anaesthetist No. 5 ignored the bradycardia and described an episode of hypotension. (Actual systolic arterial pressure values shown, at minute intervals, were: 102, 90, 88, 96, 98, 115, 101, 92, 100, 104 and 110 mm Hg.) Subsequent inspection of the anaesthetic charts confirmed that all 11 anaesthetists were unable to Table 1 Expected and observed descriptions of phase I of the simulation by anaesthetists (six anaesthetists did not describe this phase) Expected There was a gradual decline in heart rate from 80 to 35 beat min1 over the first 15 min, otherwise the patient was stable. Anaesthetist Reported events (No.) (1) (2) (3) (4) (5) (6-11) “… gradual decline in HR noted No change in ECG. No change in BP. No cause evident.” “One episode of bradycardia during GA, responded to atropine (300 g) otherwise CVS O2 Sats 996%.” “Approximately 40 min into the procedure, which was otherwise unremarkable other than bradycardia (sinus) to 42 beat min1. RX atropine.” “The first 30 min of anaesthesia and surgery was stable. The “patient” then developed a bradycardia, with the heart rate decreasing to about 30 beat min1. Arterial blood pressure was reduced but still maintained within acceptable limits.” “Initially everything was fine, then BP started to slowly drop from 120 to 90 systolic. Responded to Hartmann’s 1L.” No comment made about bradycardia Inaccurate reporting of simulated critical anaesthetic incidents 639 Table 2 Expected and observed descriptions of phase two of the simulation by anaesthetists Expected After a test dose of flucloxacillin, patient developed bronchospasm. Arterial pressure was initially normal but decreased rapidly after 2 min. Patient responded to i.v. and adrenaline. Anaesthetist Reported events (1) (2) (3) (4) (5) (6) (7) (8) (9) Figure 1 Actual and recorded heart rates for anaesthetists Nos 9, 10 and 11. record accurately the simulated events or their time scale. Figure 1 shows examples of actual heart rates plotted on the same scale as those recorded by the anaesthetist. Overall, subjects tended to record an initial stable period, followed by a brief bradycardia, rather than a gradually evolving bradycardia. The anaesthetists’ charts often showed a time scale twice as long as the real events. (10) (11) “following i.v. fluclox, Paw increased such that the Manley was ineffective. A diagnosis of anaphylaxis was made, Fi O2 to 100%. IPPV by bag. Auscultation, widespread wheeze bilat. Adrenaline 100 g given i.v. with quick result. As compliance improved m 9 Manley. 1% ISO throughout. No desaturation. No hypotension.” “On administering flucloxacillin 50 mg during CVS collapse, airway (P) increasedSats decreased to 91%.” “I was then asked to administer flucloxacillin which I did i.v. Over the course of the next few minutes, the ‘patient’ dropped his SaO2 % to 92–93%, developed bilateral bronchospasm, and also progressively dropped his blood pressure.” “‘patient’’ developed bronchospasm hypotension in response to i.v. flucloxacillin.” “Asked to give antibiotic. No Hx of allergy. Diluted flucloxacillin 1 g in 10 ml. 2 ml test dose— bronchospasm, hypotension.” “Immediately after flucloxacillin solution 1 ml— tight bronchospasm with ventilator unable to deliver volume ... Minimum SaO2 92%, ETCO2 to 3.3 kPa. BP always strongly palpable—minimum recorded 89/56 (although BP machine malfunctioned for 2 min during event)” “Flucloxacillin 1 g i.v. Events: (chronologically) (1) bronchospasm, (2) no BP recorded, (3) back to normality” “After giving test dose of fluclox. ‘patient’ became difficult to ventilate and was wheezy with very little air entry suggesting severe bronchospasm.” “Test dose of 50 g given. ‘Patient’ developed bronchospam evident by increased Paw and decreased AE within 30 s. No response to salbutamol via ETT. Subsequent hypotension (no BP on Dinamap) but faint carotid pulse.” “1-ml test dose given 9 bronchospasm 9 hard to ventilate with halothane, i.v. salbutamol adrenaline fluids.” “Bronchospasm ‘anaphylactic’ hypotension 2⬚ to i.v. fluclox ... Min BP 52/min SpO2 92%” Table 3 Assessment of realism by anaesthetists, using visual analogue scale (0unrealistic, 10realistic) Median Range score RESPONSES TO SIMULATED BRONCHOSPASM All 11 anaesthetists described the second phase of the simulation, but there were significant differences between actual and recorded events. In their critical incident reports (table 2), none of the anaesthetists commented that arterial pressure had been normal for 2 min after bronchospasm had started. Only two anaesthetists (Nos 7 and 9) correctly described the actual sequence of events. Analysis of the videotapes however, showed that most of the anaesthetists had looked at the arterial pressure reading soon after bronchospasm had started. Seven of the 11 anaesthetists either stated that arterial pressure was normal and/or increased the inspired concentration of volatile agent, indicating that most were aware of the initially normal arterial pressure. For example, anaesthetist No. 2 How do you rate the realism of the simulations? 8 How do you rate the realism of your responses to the simulation? 7 (6–8) (5–8) stated “Sats OK, BP OK” just before starting to treat the bronchospasm with aminophylline. Yet the same anaesthetist described, in the written account, the hypotension (“CVS collapse”) as occurring before or at least at the same time as the bronchospasm. The answers to the questionnaire (table 3) showed that the anaesthetists thought that the simulations were realistic. Discussion This study has shown that a group of trainee 640 anaesthetists treated a simulated “patient” in a realistic manner. However, their written accounts of events were influenced heavily by their own final diagnosis, with events being altered or omitted to fit with this diagnosis. SIMULATED BRADYCARDIA The results from phase 1 showed a tendency for the anaesthetists to ignore any events which they perceived as not being directly related to their “main” problem, in that seven of 11 did not report that bradycardia had occurred. They also showed a tendency to “normalize” events. A gradual decrease in heart rate is unusual, but this was mentioned by only one anaesthetist who wrote, “No cause evident”. Anaesthetists Nos 2, 3 and 4 mentioned that bradycardia had occurred. The implication is that the anaesthetists were not able to remember individual values for each of the variables shown, but were recording events as complete episodes. This view is reinforced by the recorded values on the anaesthetic charts (fig. 1) which tended to show an initial period of normal heart rate followed by sudden bradycardia. In other words, the first phase of the simulation had been remembered as two discrete episodes, an initial episode of normality, followed by an episode of bradycardia. When writing about the events afterwards, values for heart rate and arterial pressure were obtained from their own mental models of those events. This may also explain why anaesthetist No. 5 reported a single reading of systolic arterial pressure of 88 mm Hg as an episode of hypotension. This was only a small decrease from the real initial value of 102 mm Hg. However, he reported an initial value of 120 mm Hg, that is a “normal” arterial pressure, which would make a subsequent systolic pressure of 88 mm Hg more significant. The next reading of a systolic arterial pressure of 115 mm Hg may have indicated that the i.v. fluid had been successful in treating the “hypotension”, reinforcing the diagnosis. A later value of 90 mm Hg was then ignored. SIMULATED BRONCHOSPASM During phase 2 all anaesthetists responded appropriately to the increased airways resistance and appeared to be aware of the normal arterial pressure initially and to make a diagnosis of isolated bronchospasm. Subsequently, when arterial pressure began to decrease, the anaesthetists seemed to change the diagnosis from bronchospasm to a generalized anaphylactic reaction. Then, when their treatment began to be effective and heart rate and arterial pressure began to return to normal, the new diagnosis of anaphylaxis with cardiovascular collapse was reinforced. Therefore, the episode was remembered as an episode of anaphylaxis and any inconsistent information was subconsciously suppressed, or not remembered. Within the specialty of anaesthesia, an accurate record is of prime importance and errors have tended previously to be described in terms of inattention or misdiagnosis. This study tested British Journal of Anaesthesia anaesthetists during daylight hours, without the distractions of bleeps, telephone calls and audit forms. The short period of the study also implied that boredom was unlikely to have been a factor. All “patients” were treated correctly and the anaesthetists appeared to have few difficulties in understanding the data presented by the simulator. Although an anaesthetist “taking over” a case half way through may not be regarded as “best practice”, trainee anaesthetists are often asked to take over the care of patients for short periods. This is usually regarded as acceptable provided the incoming anaesthetist is briefed correctly, as was the case in this study. The conclusion is therefore that the errors observed were caused by problems with the recall of events by the anaesthetists. The problems described here have been observed in other fields such as aviation and the nuclear power industry and have been the subject of considerable research.2 3 In particular, our limited capacity to retain numerous discrete elements of data4 and the tendency of subjects to act on the present value of a variable rather than the trend5 have been demonstrated. The observation that memories are stored as typical “patterns” or “schemata” has been described as far back as 1934.6 The relevance to anaesthetic practice was described by Allnutt in 1987, “Thus, with the highest integrity, we soon start reporting not what happened, but what must have happened” (authors italics).7 These problems have several important consequences for those analysing critical incidents. First, if anaesthetists are taught to deal with a list of well defined emergencies, with their attendant signs, symptoms and solutions, it is likely that they may try subconsciously to make any real emergency “fit” one of this list. Any record of the event may then resemble the textbook example more than the real events, with the consequence that vital details may be obscured. In particular, complications or associations which have not been recognized previously may continue or go unreported. Second, any aspect of the real event which was unusual or incompatible with the diagnosis made tends to be ignored or reduced in value. Any details recorded tend to support the final diagnosis. This study can be criticized for not using real cases and for the analysed records not haying been made contemporaneously. However, the anaesthetists reacted appropriately to the simulations and rated the experience as realistic. This is reinforced by the observation that one anaesthetist administered a dose of neuromuscular blocker to stop a plastic resuscitation manikin from “coughing”. Although the records were made immediately after the event and not at the same time, it is unlikely that during a real critical incident anaesthetists would have the time to fill in an anaesthetic chart or to review trends on a monitor. Therefore, during a critical incident, when decisions have to be made, anaesthetists have to rely on their own memory for details of recent events, and that is what has been tested here. Finally, our knowledge of anaesthetic emergencies, their diagnosis and treatment has largely been built up from analysis of critical incident forms. This study Inaccurate reporting of simulated critical anaesthetic incidents indicates that the information derived from that source may not reflect actual events. Although some anaesthetic machines are capable of providing detailed case records, monitors may not provide accurate readings during a sudden catastrophic event such as anaphylaxis. In addition, they do not record all the actions of the anaesthetist. Cases where the automated record differed substantially from the events perceived by the anaesthetist may cause difficulties for both trainers and those seeking to defend against litigation. Automated recording of monitoring and videotaping of the anaesthetic would seem to provide the best solution, but this is unlikely to receive widespread acceptance and has significant cost implications. Overall, the most immediate solution is for those involved with the analysis of critical incidents and in the debriefing of the participants to be aware that written and verbal accounts of events may be inaccurate. In addition, trainers need to ensure that anaesthetists are aware of the limitations of their perception. In particular, it is vital for 641 anaesthetists to constantly re-check diagnoses and assumptions as a critical event develops. References 1. Byrne AJ, Hilton PJ, Lunn J. Basic simulations in anaesthesia. A pilot study of the ACCESS system. Anaesthesia 1994; 49: 376–381. 2. Reason J. Human Error. Cambridge: Cambridge University Press, 1990. 3. Groen GJ. Patel VL. Medical problem solving: some questionable assumptions. Medical Education 1985; 19: 95–100. 4. Miller GA. The magical number seven plus or minus two: Some limits of our capacity for processing information. Psychological Review 1956; 63: 81–97. 5. Dorner D. On the difficulties people have in dealing with complexity. In: Rasmussen J, Duncan K, Leplat J, eds. New Technology and Human Error. Philadelphia: John Wiley and Sons, 1987; 97–109. 6. Bartlett FC. Remembering: A Study in Experimental and Social Psychology. Cambridge: Cambridge University Press, 1932. 7. Allnut MF. Human factors in accidents. British Journal of Anaesthesia 1987; 59: 856–864.
© Copyright 2025 Paperzz