1 Mortality associated with anaesthesia TOM PEDERSEN
The evaluation of perioperative complications that result in death is important for understanding and improving the quality of anaesthesia. Recent studies estimate that anaesthesia-related deaths range from 0.1 to 3 per 10 000 administrations of anaesthesia (Table 1). The methods used to gather information about the safety of anaesthesia and to establish the risk of mortality have included anecdotal tales, in-hospital audit and peer reviews, critical incident techniques, reports to medical defence societies, retrospective studies, reviews of specific problems and prospective studies. It is difficult to m a k e accurate comparisons a m o n g studies because of the differences in research and data collection methods. The assessment of anaesthetic risk is based on rather indirect approaches, taking into consideration the physical status of the patient, as well as surgical and anaesthetic factors, all of which m a y contribute either independently or jointly to adverse outcome following administration of anaesthesia. Variations in the definitions of perioperative death and differences in anaesthetic practice a m o n g countries also affect the reported incidence of risk. With these considerations in mind, the objectives of this chapter are to review the recent mortality studies in anaesthesia in order to identify the c o m m o n e s t reasons for death associated with anaesthesia, to explain why some recent mortality studies have indicated that the mortality rate m a y be increasing again, and to identify possible methods of prevention.
Table 1. Summaryof mortalityassociated with anaesthesiain major studies of the past 15 years. Reference Outcome period Incidenceof mortality Lunn and Mushin (1982) 6 days 1:10 000 Keenan and Boyan (1985) 1:10 000 Tiret et al (1986) 24 hours 1:13 207 Holland (1987) 24 hours 1:26 000 Buck et al (1987) 30 days 1:185 056 Forrest et al (1990) 7 days 1:10 000 Harrison (1990) 24 hours 1:14 000 Pedersen (1994) 30 days 1:2500 BailliOre's Clinical Anaesthesiology-
Vol. 10, No. 2, July 1996 ISBN 0-7020-2101-6 0950-3501/96/020237 + 14 $12.00/00
237 Copyright © 1996, by Ballli~reTindall All rights of reproduction in any form reserved
238
T. PEDERSEN
DEFINITION One of the main differences among mortality studies is the definition of anaesthetic death. Most studies include some variation in the number of anaesthetic administrations over the period of time and in the population under study as a denominator value (Lnnn, 1986). Identification of numerator values is a more complicated matter, which also differs among the studies. Most studies state a period of time in which a death had to occur in order to merit further consideration. Variations in the definition of mortality associated with anaesthesia (within 24 hours of surgery to 30 days after operation), in patient population, and differences in anaesthetic practice among countries affect the reported incidence. Mortality associated with anaesthesia may be defined as any death occurring during or following anaesthesia, apparently caused by anaesthesia or some aspect of anaesthetic management. COMMON CAUSES OF MORTALITY ASSOCIATED WITH ANAESTHESIA The factors that have contributed to anaesthetic mortality (Table 2) have remained relatively constant over the years, although their relative importance differs in different studies. Death as a complication of anaesthesia is usually an event that follows loss of clinical control of the patient's
Table 2. Factors commonly reported as contributing to anaesthetic mortality. During anaesthesia Circulatory instability Hypovolaemia Hypotension Severe cardiac arrhythmia Hypoxaemia and hypoventilation Oesophageal intubation Undetected ventilator disconnection Inability to ventilate after anaesthesia induction Inadequate inspired oxygen concentration Pulmonary aspiration of stomach contents Drug overdose and wrong choice of agents Anaphylactoid shock Intravenous injection of local anaesthetic agents Total spinal block Inadequate monitoring (late detection of hypoxaemia) After anaesthesia Inadequate postoperative monitoring and management Ventilatory failure Hypoxaemia Hypoventilation Circulatory failure Postanaesthetic cardiac failure with increased afterload
MORTALITY
ASSOCIATED
WITH ANAESTHESIA
239
physiological homoeostasis and the functional failure of interrelated systems, commonly the respiratory and circulatory systems (Harrison, 1990). In assessing these, consideration must be given not only to the ways in which the known actions of anaesthetic drugs or errors and misjudgements in their administration may have fatal consequences, but also to the ways in which errors of commission or omission in the wider field of responsibility of the anaesthetist may cause death. A broad classification of the mechanism or cause of death associated with anaesthesia was adopted by Harrison. 1. Failure to control circulatory homoeostasis. 2. Failure to control respiratory homoeostasis. 3. Complications of regional anaesthesia. Those causes of death allocated to failure in control of respiratory and circulatory homoeostasis, the major groups, were in turn subdivided in terms of the mechanisms or clinical failures responsible for them. Harrison (1990) found that circulatory failure included failure in the management of blood volume (15%), failure in control of arrhythmia, and drug-related or vagal causes (15%). Failure in control of respiratory homoeostasis included complications of tracheal intubation (18%), postoperative hypoventilation due to relaxant or narcotic agent (15%), bronchial obstruction (6%), failure in management of ventilation during anaesthesia (5%), and vomiting and regurgitation (3%). The process of injury, which starts with some incident and ends in death, is a dynamic process that could be averted by correct intervention at any stage, as stressed by Harrison (1990). This implies that most fatal incidents involve multiple errors, which in turn activate the composite mechanisms that ultimately result in death. Holland (1987) reported an average of 4.3 errors per death in the findings of the New South Wales Special Committee on Death under Anaesthesia. Thus, there may be a problem in deciding to which of the multiple errors the death should be attributed for classification purposes.
Failures in control of circulatory homoeostasis Death due to the anaesthetist's failure to control the patient's circulatory homoeostasis is responsible for about one third of deaths associated with anaesthesia, and can be classified into two groups: (1) those that follow from failure to maintain a normal circulating blood volume; and (2) cardiac arrest from causes other than anoxia or ischaemia (e.g. failure of arrhythmia control). Hypovolaemia is included in the category of circulatory failure. In Harrison's report, clinical factors associated with this situation were: 1. 2.
In more than two thirds of the cases there was evidence of uncorrected preoperative hypovolaemia. Seventy-five per cent of the patients were elderly and showed evidence of degenerative vascular disease and ischaemic heart disease.
240 3. 4.
T. PEDERSEN
In 50% of cases, profound hypotension closely followed the administration of thiopentone for induction of anaesthesia. A subset of patients warrants special mention as it highlights the grave risk that anaesthesia poses to a common group of surgical patients, and to those who present for lower-limb amputation because of diabetic gangrene. These patients are elderly, with arteriosclerosis and diabetes mellitus, and possibly suffering from ischaemic heart disease.
Cardiac arrest may also be caused by administration of drugs: 1. 2. 3.
Suxamethonium can cause severe bradycardia proceeding to asystole (1:1000) (Pedersen, 1994), particularly in the presence of severe acidosis or endotoxaemia. Halothane can cause ventricular fibrillation and halothane hepatitis, but these events are so rare that they seldom appear, even in large studies. Neostigmine can cause bradycardia going on to asystole.
Harrison (1990) also found that cardiac arrest can be associated directly with severe vagal stimulation and that inadvertent transfusion hypothermia was associated with the onset of ventricular fibrillation. Failure in control of respiratory homoeostasis Failure in airway management was the most common general cause of death associated with anaesthesia (57%) in Harrison's study (1990). Unrecognized oesophageal intubation, undetected ventilator disconnection and an inability to ventilate after induction of unconsciousness and paralysis are all examples of inadvertent failure in control of respiratory homoeostasis. Together, they are the commonest causes of hypoxaemia in all studies cited. Hypoxaemia is unquestionably the commonest cause of anaesthetic disaster and death. Every study of anaesthetic mortality supports this contention (Keenan and Boyan, 1985; Caplan et al, 1988; Pedersen and Johansen, 1989; Williamsen et al, 1993). Collections of anaesthetic mishaps, including those involving litigation, indicate that from one third to two thirds of those that end in death are the result of hypoxaemia (Utting et al, 1979; Holland, 1987). In the study of Harrison (1990), nine patients (7% of anaesthetic deaths) died as a result of hypoxaemia that arose from tracheal tube obstruction due to kinking of the tube or cuff herniation. The latter complication is particularly treacherous in that it can arise some time after intubation with few or no warning signs. Four patients (3%) died from hypoxaemia and vagally induced cardiac arrest when acute massive bronchospasm and expiratory muscle spasm (induced by attempts at iutubation when anaesthesia was too light, laryngeal topical analgesia insufficient and muscle relaxation inadequate) rendered pulmonary ventilation impossible. Closely related to these tracheal tube complications was a group of eight deaths (6%) associated with hypoxaemia consequent on a failure to clear obstruction of the lower airway by pus, blood or secretions from pulmonary or oesophageal disease.
MORTALITY
ASSOCIATED WITH ANAESTHESIA
241
Despite the introduction of monitoring by pulse oximetry which seems to reduce the extent of hypoxaemia (M¢ller et al, 1992), the advent of other monitoring devices designed to enable the detection and correction of potentially harmful events, and several changes in patient care, no reduction in the number of respiratory complications or in mortality rate was found in the prospective, randomized, clinical Danish study (MOller et al, 1993). This study was not able to confirm the findings of the American Society of Anesthesiologists (ASA) closed claims analysis or other retrospective safety studies, because these studies concern only serious events, nearly all resulting in serious disability or death (Caplan et al, 1990). In the closed claims analysis of adverse respiratory events in anaesthesia, reviewers judged that better monitoring would have prevented the adverse outcome in 72% of claims. The reviewers chose pulse oximetry and capnography or both of these procedures in 98% of cases (Caplan et al, 1990). In the general analysis of the role of monitoring techniques in prevention of anaesthetic mishaps, nearly 60% of deaths were considered preventable by application of additional monitors. The Danish prospective randomized study (M¢ller, 1994) suggested that pulse oximetry affects patient care but did not confirm the hypothesis that general postoperative complications, including overall mortality, were decreased by pulse oximetry. However, we still lack definitive knowledge of the impact of perioperative pulse oximetry monitoring on severe complications and anaesthetic-related death. The second most common cause of anaesthetic death is failure to ensure adequate pulmonary ventilation (about 20% of deaths associated with anaesthesia). The majority of these events occur at the termination of anaesthesia or in the immediate or short-term post-anaesthetic recovery period (Harrison, 1990; Pedersen, 1994). In the latter study, many of these episodes of hypoventilation were related to inadequate reversal of muscle relaxants with consequent hypoventilation that was mismanaged. Common contributors to this situation were abdominal distension from intestinal obstruction, chronic obstructive lung disease, obesity or a combination of all three. The most common fault in clinical management was the adoption for too long after attempted reversal of neuromuscular block of a 'wait and see' policy, and not using, for example, tactile evaluation of the response to train-of-four nerve stimulation to assess postoperative residual neuromuscular blockade. Aspiration of stomach contents has been cited as a common cause of disaster and anaesthetic death. Aspiration caused about 8% of anaesthetic deaths in studies of anaesthesia in the years 1945-1954 (Hingson et al, 1956), 4% in a study from 1967 to 1976 (Harrison, 1978) and 3% in the recent study from Harrison (1990). Many of these deaths occurred in obstetric patients and the decreasing incidence may be due in part to increased utilization of regional anaesthesia for parturients. Death due to equipment misuse or technical failure
Modern alarm systems and monitoring standards should draw the anaesthetist's attention to a hazardous situation before it becomes clinically
242
T. PEDERSEN
irretrievable. Further, such events usually provoke changes in equipment design, so that these particular errors are unlikely to be repeated. In the analysis over a 30-year period from 1956 to 1987, five cases were found in which equipment misuse resulted in death (Harrison, 1990). In the recent prospective studies in Denmark (1986-1987 and 1990-1991) no case of equipment failure was recorded (Pedersen and Johansen, 1989; M¢ller et al, 1993). C O M M O N CAUSES OF OVERALL MORTALITY ASSOCIATED WITH ANAESTHESIA AND SURGERY Mortality rate and risk indicators
The risk of dying in non-cardiac surgery is about 0.05% during anaesthesia, 0.1% during the recovery period (the first 24 hours after operation) and 0.6% in the first 6 days; the overall mortality rate in hospital is about 1% (Lunn and Mushin, 1982; Pedersen et al, 1990). Elderly patients are at highest risk of death. Farrow et al (1982) found that the relative risk in relation to anaesthesia and surgery of a concurrent disease such as ischaemic heart disease decreased with advancing age. There is a lower percentage of elective surgery in the elderly and the surgical procedure is often extensive, which carries a higher risk of death. However, it is also a fact that organ function is reduced with age. It is therefore reasonable to assume that age itself carries a risk. A high mortality rate is expected in patients with pre-existing cardiac disease, especially those undergoing major surgery, and in the presence of clinical signs of chronic heart failure, recent acute myocardial infarction or preoperative hypotension (Table 3). From a national register of hospitalized patients, Dirksen and Kj¢ller (1988) selected patients with a diagnosis of myocardial infarction within 1 year, who were scheduled for either appendicectomy or surgery for proximal femur fracture. The mortality rates within 1 month were 9% and 16% respectively. They found that a history of chronic heart failure proved to be the dominating risk factor, while ischaemic heart disease had no independent association with mortality related to anaesthesia and surgery. It is likely that the degree of myocardial damage and ventricular impairment quantified by echocardiography or radionuclide cardiography is a better predictor of fatal complications than simply a history of angina pectoris. In the Copenhagen study (Pedersen, 1994) it was found that 67% of patients who developed myocardial infarction following anaesthesia died in the postoperative period. Before operation 75% had clinical symptoms of heart failure and 50% experienced episodes of severe cardiovascular failure during anaesthesia. It was noteworthy that 50% of the patients with myocardial infarction who died received regional anaesthesia, but nevertheless experienced cardiovascular dysfunction during operation. This suggests that caution should be exercised when applying spinal anaesthesia to patients with cardiac disease.
243
MORTALITY ASSOCIATED WITH ANAESTHESIA
Table 3. Risk factors in relation to incidence of mortality in hospital. No. of anaesthetics Risk factor
Mortality in hospital
(N= 7306)*
%
Odds ratio
4587 2634
0.7 2.21.
3.2
3965 2043 886 293 103
0.3 1.8 2.91. 5.81, 2.9
2.5
125 26 199 380 127 201 153 141 34 1257 242 2454
4.01, 7.71. 9.01. 1.3 9.41, 5.01 5,9I" 2.11. 2.9 1.1 5.01. 2.8I
9.9 1.1 9.8 4.7 5.2 1.8 2.4 1.0 5.4 3.2
1774 4621 470 162 4916 2113
0.1 1,3 3.21, 4.91, 0.3 3.11,
4,9
Sex
Female Male Age (years) <50 50-69 70-79 -> 80 Ischaemic heart disease Myocardial infarction > 1 year _< 1 year Chronic heart failure Hypertension Hypotension (systolic blood pressure < 90 m m Hg) Chronic obstructive lung disease Renal failure Diabetes mellitus Neurological disease Cancer Abdominal Emergency surgery Duration of anaesthesia (minutes) < 30 30-179 180-299 > 300 Minor surgery Major surgery
Total
1.2
* The analysis shows the maximum number of patients in the various groups. "1"P < 0.05 v e r s u s rest of total incidence. Reproduced from Pedersen (1994, Danish Medical Bulletin 41: 319-331) with permission.
The prognosis seems to be better when cardiopulmonary complications occur in the intraoperative period (about 5% of lethal complications) rather than after operation (about 15-20%) (Pedersen et al, 1990). These findings are similar to those reported by Tiret and colleagues (1986), who found a twofold to threefold increase in the risk of death when the complications occurred in the postoperative period. The duration of surgery and anaesthesia is known to influence the mortality risk, but the risk-predictive value of this factor is questionable. It may rather reflect the severity of the underlying surgical disease and the extent of the surgical treatment that is needed. Risk factors that influence fatal complications in anaesthesia are numerous, and any predictions based on preoperative clinical status alone must be incomplete. It has been suggested, however, that risk assessment
244
T. PEDERSEN
based on detailed evaluation of different individual risk variables summarized into a multifactorial risk index allows an individualized estimation of risk and optimal management of anaesthesia. The overall mortality rate is strongly associated with elderly patients (over 70 years of age), patients with preoperative hypotension (systolic blood pressure below 90 mm Hg), clinical signs of chronic heart failure, as well as emergency procedures involving major abdominal surgery. HAS THE INCIDENCE OF MORTALITY ASSOCIATED WITH ANAESTHESIA CHANGED IN RECENT YEARS? Review of the major anaesthetic risk surveys of recent years reveals considerable variation in the incidence of mortality associated with anaesthesia (Table 1). The risk of death associated with anaesthesia decreased from 1 in 2680 to 1 in 10 000 in the period from 1954 to 1982 (Beecher and Todd, 1954; Lunn and Mushin, 1982). However, in the report of a confidential enquiry into perioperative deaths which took place in 1985, only three deaths in 555 258 anaesthetic administrations were due solely to anaesthesia, an incidence of one death in 185 056 anaesthetic administrations (Buck et al, 1987). Zeitlin (1989) reported a possible 13fold decrease in death rate related to general anaesthesia in the period 1977-1984 compared with 1955-1964. There have also been studies of cardiac arrest during anaesthesia. The Medical College of Virginia study reported originally in 1985 (Keenan and Boyan, 1985) was extended to include two full decades of experience, each with a total of more than 100 000 anaesthetic administrations (Keenan and B oyan, 1991). The first decade, 1969-1978, pre-dated the safety initiatives of the second decade, 1979-1988, during which pulse oximetry and capnography were introduced. The intraoperative anaesthetic cardiac arrest rate fell by 50%, from 2 per 10 000 administrations of anaesthetics in the first decade, to 1 per 10 000 in the second. A study of anaesthetic cardiac arrest reported from the Netherlands found 1.3 arrests per 10000 anaesthetic administrations; one patient per 16 250 anaesthetics could not be resuscitated successfully (Chopra et al, 1990). This rate is comparable to that of the second decade of the Keenan study in almost the same time period, and both are lower than the rates from earlier studies. These data suggest that the incidence of anaesthetic cardiac arrest decreased significantly in the 1980s. In contrast, two recent studies of anaesthetic cardiac arrest and mortality reported from Denmark found 15 cardiac arrests per 20 802 non-cardiac anaesthetics administered, and death was considered to be due partly or totally to anaesthesia in approximately 1 per 2500-3000 anaesthetics (M¢ller, 1994; Pedersen, 1994). This rate is comparable to that in a French survey study of 198 103 anaesthetics, where the incidence of death due partly or totally to anaesthesia was 1 per 2387, and death due totally to anaesthesia had an incidence of I per 13 207 (Tiret et al, 1986). These findings in the 1990s show that the incidence of anaesthetic mortality seems to be at the same level as that in the mid-1980s. One may
MORTALITY
ASSOCIATED
WITH ANAESTHESIA
245
say, like Keats (1990): 'wishful thinking is the only basis for believing that anaesthetic mortality today is less than it was 40 years ago'. HAVE THE CLINICAL CAUSES OF MORTALITY ASSOCIATED WITH ANAESTHESIA CHANGED IN THE PAST 25 YEARS?
Keats (1990) stated that the total risk of death attributable directly to anaesthesia had decreased during the past 25 years. However, he identified some major limitations of the various anaesthesia mortality studies. In particular, the studies used non-parallel methodologies, which makes comparison of their results difficult. Holland (1987) found that it appeared to be about five times safer to undergo anaesthesia in 1987 than in 1960 (number of administrations per death: 1 in 26 000 versus 1 in 5500). The reduction in risk was even greater for a young and otherwise healthy person. One of the most spectacular changes during that period of time was the increase in the number of trained anaesthetists (Fellows of the Faculty of Anaesthetists of the Royal Australian College of Surgeons) serving the population, which increased from 106 to 1115. The changes in aetiological pattern of causes of mortality associated with anaesthesia has been described by Harrison (1990). There was a progressive increase in the proportion of deaths due to failure in the control of respiratory homoeostasis accompanied by a reciprocal decrease in those due to failure in the control of circulatory homoeostasis. The four most common clinical causes of death reported (failure in the management of airway, ventilation, blood volume and arrhythmia) showed that failure in airway management, of which the majority involved complications of endotracheal intubation, was responsible for the increased proportion of respiratory deaths. The closed claims study in the United States (Cheney et al, 1991) also found a higher incidence of failure related to the respiratory (37%) than to the cardiovascular (6%) system in patients with a serious adverse outcome (death or cerebral damage). Harrison (1990) suggested that errors in airway management are now more easily perpetuated than those involving the cardiovascular system. An explanation for this may lie in the fact that, whereas endotracheal intubation and mismanagement of the airway involve physical skills, manual dexterity and a degree of informed clinical judgement, control of circulatory homoeostasis depends to a large extent on education, anaesthetic methods and monitoring systems that have improved in sophistication, acceptability and availability in recent years. The Confidential Enquiry into Maternal Deaths
The recently published triennial Report on the Confidential Enquiry into Maternal Deaths in the United Kingdom (1994) represents the latest report of a clinical audit that has been in continuous existence since 1952. It is particularly noteworthy that there has been a sharp reduction in the
246
T. PEDERSEN
proportion of maternal deaths associated directly with anaesthesia (Table 4). However, these data do not provide a true indication of the improvement in quality of anaesthesia. As observed in 1986 (Morgan), between 1970-1972 and 1979-1981, there was a marked increase in the total number of caesarean sections and an enormous increase in the number of legal abortions. It is unlikely that this trend has continued and so the number of anaesthetic deaths as a proportion of the number of anaesthetic interventions may have declined at an even greater rate. In the last two triennia, there was a reduction in the percentage of direct anaesthetic deaths, from 4.3% to 2.7% of all maternal deaths. Table 4. Maternal deaths associated directly with anaesthesia. Reproduced from Confidential Enquiry into Maternal Deaths in the United Kingdom (1994).
England and Wales 1973-1975 1976-1978 1979-1981 1982-1984 UK 1985-1987 1988-1990
Deaths primarily related to anaesthesia
Rate of direct anaesthetic deaths per million pregnancies/maternity
Direct anaesthetic deaths as a percentage of direct maternal deaths
27 27 22 18
10.5 12.1 8.7 7.2
11.9 12.4 12.4 13.0
6 4
1.9 1.7
4.3 2.7
IS REDUCTION OR PREVENTION OF MORTALITY ASSOCIATED W I T H ANAESTHESIA POSSIBLE? Several possible mechanisms which might reduce or prevent mortality associated with anaesthesia have been proposed.
Identification of contributory factors The best chance of reducing the number of cases with fatal outcome probably lies in the identification of factors that contribute to anaesthetic mortality. Each accident may be unique in itself, but is often the end result of a chain of events, components of which may be common to a range of accidents. Gannon (1991) found that inadequate supervision emerged as a significant factor in the fatal cases, as it did in a similar review of obstetric accidents (Ennis and Vincent, 1990) and in audits of mortality related to anaesthesia (Lunn and Mushin, 1982), and surgery and anaesthesia combined (Buck et al, 1987). Criticisms were also made about inadequate preoperative assessment. In several cases, the patient was seen by an anaesthetist for the first time in the anaesthetic room. In some instances the issue of history taking was linked with a failure of communication between surgical and anaesthetic teams, in that anaesthetists were not informed
247
MORTALITY ASSOCIATED WITH ANAESTHESIA
about certain relevant aspects of the patient's history. The Confidential Enquiry into Perioperative Deaths (CEPOD) report also drew attention to the issue of communication between the anaesthetist and surgeon (Buck et al, 1987). Communication problems may also occur between medical and nursing staff (Gannon, 1991). The problems associated with history taking, assessment of patients before anaesthesia and the discharge of patients suggest a need for standardized protocols to encompass these areas. In view of the increasing amount of day-case work, which makes greater demands on junior doctors and paramedical staff, the use of standardized protocols would seem to be particularly important. Are high-risk patients more likely to be affected by errors resulting in a substantial negative outcome? A common view is that anaesthetic risk should be zero because anaesthesia is not in itself therapeutic and anaesthetic agents themselves are not lethal except when misused. However, it has been shown that less healthy patients (ASA grade 3 or worse) are more likely to be affected adversely by errors and to have a substantial negative outcome (brain damage or death); the incidence of preventable errors related to complications attributable to anaesthesia in seriously ill patients (ASA grade 3 or worse) is six times greater than that in patients of ASA grade 1 or 2 (Table 5) (Pedersen and Johansen, 1989). Surprisingly, only one third of preventable errors involved
Table 5. Anaesthesia-attributable complications for each ASA physical grade and for emergency and elective cases in relation to negative outcome and preventability (Pedersen, 1994). Complications attributable to anaesthesia (CAA)
Negative outcome (NO)
Preventable errors (PE)
NO/CAA
PE/CAA
%
N
%
N
%
N
%
N
%
N
0.68
33
0.04
2
0.27
13
6.1
2/33
39.4
13/33
0.41
10
0.16
4
0.12
3
40.0*
4/10
30.0
3/10
Healthy or moderately ill ASA 1-2 (N=6502) 0.34
22
0.00
0
0.14
9
0.0
0/22
40.9
9/22
Seriously ill ASA/> 3 (N= 804)
2.61"~
21
0.75t
6
0.87t
7
28.6t
6/21
33.3
7/21
Total
0.59
43~:
0.08
6
0.22
16
14.0
6/43
37.2
16/43
Operation Elective (N=4852) Emergency (N= 2454)
* P < 0.05 versus elective operation; t P < 0.05 versus ASA grade 1-2. $ Three patients died. Reproduced from Pedersen (1994, Danish Medical Bulletin 41: 319-331) with permission.
248
T. PEDERSEN
anaesthetists in training. The remaining incidents involved staff anaesthetists. Chopra et al (1990) also found that the majority of deaths associated with anaesthesia (61%) involved patients with ASA grades 4 or 5. None of the patients in grade 4 and 5 who developed cardiac arrest could be resuscitated successfully. There was thought to be a preventable cause for the cardiac arrest in 47% of the cases. Errors in drug administration and hypoxaemia were the most commonly associated preventable factors.
Has pulse oximetry monitoring contributed to prevention of anaesthetic mortality? The Danish study (M¢ller, 1994) has not answered the question of whether monitoring with pulse oximetry reduces the incidence of catastrophic events resulting in serious disability or death. To answer these questions and other problems concerning anaesthetic safety, improved methods are needed for evaluation of new standards and monitoring equipment.
Can anaesthetic-related maternal mortality be reduced? Anaesthesia continues to be a prominent cause of maternal mortality, accounting for approximately 10% of pregnancy-related deaths in some countries. At least 90% of maternal deaths are attributable to general anaesthesia, primarily as a result of failed endotracheal intubation or pulmonary aspiration of gastric contents, despite the fact that more than 50% of caesarean sections are performed under regional anaesthesia. May and Greiss (1989) reported comparable statistics for anaesthetic-related deaths, all of which were attributable to the use of general anaesthesia, and concluded that we have reached an irreducible minimum in anaestheticrelated deaths with more complications occurring now from morbid obesity than from the gravid state. However, the increased use of regional anaesthesia, which rarely results in maternal mortality, will almost certainly reduce the number of deaths from anaesthesia because regional anaesthesia does not impair the parturient's ability to protect her airway from pulmonary aspiration of gastric contents and obviates the need for endotracheal intubation for which failure to accomplish the procedure, in obstetrics, has been reported in 1 in 300 attempts, with difficulty encountered in 5% of cases (Morgan, 1986). Obese individuals are exposed to much greater risks than non-obese patients during anaesthesia because of more difficulty with tracheal intubation, higher gastric volume, lower gastric pH and diminished barrier pressure (lower oesophageal sphincter tone minus intragastric pressure) compared with normal patients. The fact that three of the four anaesthetic-related deaths reported by May and Greiss (1989) resulted from difficult endotracheal intubation in markedly obese individuals suggests that the obese parturient is especially suited for regional anaesthesia, and anaesthetic-related maternal mortality rates can be improved.
MORTALITY ASSOCIATED WITH ANAESTHESIA
249
SUMMARY
In the mid-1990s, most statistics are still close to the generally accepted modern risk of one death due to anaesthesia per 10000 administered anaesthetics. However, anaesthetists are faced with an increasing number of patients who are at high risk and who undergo increasingly extensive and prolonged surgical interventions. With the increasing complexity of anaesthetic care, the increasingly complex equipment and increasing demands on anaesthetists, an increase in morbidity and death rates may be expected. There are therefore strong indications to develop and use appropriate protocols for preoperative assessment, perioperative anaesthetic technique and monitoring, and postoperative care and treatment of complications. Problems with training and supervision continue to appear as factors associated with both near misses and death, and concerns about training and supervision continue to be expressed. The Colleges of Surgeons in England and Denmark have recently emphasized the need for consultants to participate fully in the training and supervision of trainees. For some time now, decision theory (Dowie and Elstein, 1988) has been applied to general clinical decision making and it may be relevant to apply this technique to an analysis of factors involved in accidents. The finding that anaesthetists of consultant grade were principally involved in many of the cases reported in the studies of Gannon (1991) and Pedersen and Johansen (1989) suggests that decision making by senior doctors might appropriately be studied in this way. Further developments of new anaesthetic techniques and perioperative monitoring equipment, together with increased focus on training and supervision, have the potential to improve anaesthetic care and reduce the incidence of fatal complications in anaesthesia.
REFERENCES Beecher HK & Todd DP (1954) A study of deaths associated with anaesthesia and surgery based on a study of 599 548 anesthesias in 10 institutions 1948-52 inclusive. Annals of Surgery 140: 2-35. Buck N, Devlin HB & Lunn JN (1987) Report on the Confidential Enquiry into Perioperative Deaths. Nuffield Provincial Hospitals Trust, London. London: Kings Fund Publishing House. Caplan RA, Ward RJ, Posner K et al (1988) Unexpected cardiac arrest during spinal anesthesia: a closed claims analysis of predisposing factors. Anesthesiology 68:5-11. Caplan RA, Posner KL, Ward RJ et al (1990) Adverse respiratory events in anesthesia: a closed claims analysis. Anesthesiology 72: 828-833. Cheney FW, Posner KL & Caplan RA (1991) Adverse respiratory events infrequently leading to malpractice suits. A closed claims analysis. Anesthesiology 75: 932-939. Chopra V, Bovill JG & Spierdijk J (1990) Accidents, near accidents and complications during anaesthesia. Anaesthesia 45: 3-6. Dirksen A & Kj¢ller E (1988) Cardiac predictors of death after non-cardiac surgery evaluated by intention to treat. British Medical Journal 297:1011-1013. Dowie J & Elstein A (1988) Professional Judgement: A Reader in Clinical Decision Making. Cambridge: Cambridge University Press. Ennis M & Vincent CA (1990) Obstetric accidents: a review of 64 cases. British Medical Journal 300: 1365-1367.
250
T. PEDERSEN
Farrow SC, Fowkes FGR, Lunn JN et al (1982) Epidemiology in anaesthesia. II: Factors affecting mortality in hospital. British Journal of Anaesthesia 54:811-817. Forrest JB, Cahalan MK, Rehder K et al (1990) Multicenter study of general anesthesia. II. Results. Anesthesiology 72: 262-268. Gannon K (1991) Mortality associated with anaesthesia. Anaesthesia 46: 962-966. Harrison GG (1978) Death attributable to anaesthesia. British Journal of Anaesthesia 50: 1041-1046. Harrison GG (1990) Death due to anaesthesia at Groote Schuur Hospital, Cape Town--1956-1987. South African Medical Journal 77: 416-421. Hingson RA, Holden WD & Barnes AC (1956) Mechanisms involved in anesthetic deaths. A survey of operating room and obstetric delivery room related mortality in the University Hospital of Cleveland, 1945-1955. New York State Journal of Medicine 56: 230-236. Holland R (1987) Anaesthetic mortality in New South Wales. British Journal of Anaesthesia 59: 834--841. Keats AS (1990) Anesthesia mortality in perspective. Anesthesia and Analgesia 71:113-119. Keenan RL & Boyan CP (1985) Cardiac arrest due to anesthesia. A study of incidence and causes. Journal of the American Medical Association 253: 2373-2377. Keenan RL & Boyan CP ( 1991) Decreasing frequency of anesthetic cardiac arrest. Journal of Clinical Anesthesia 3: 354-357. Lunn JN (1986) Epidemiology in Anaesthesia. London: Edward Arnold. Lunn JN & Mushin WW (1982) Mortality Associated With Anaesthesia. Oxford: Nuffield Provincial Hospital Trust. May WJ & Greiss FC Jr (1989) Maternal mortality in North Carolina: a forty-year experience. American Journal of Obstetrics and Gynecology 161: 555-561. M¢ller JT (1994) Anesthesia related hypoxemia--the effect of pulse oximetry monitoring on perioperative events and postoperative complications. Thesis, L~egeforeningens Forlag, Copenhagen. M¢ller JT, Jensen PF, Johannessen NW et al (1992) Hypoxaemia is reduced by pulse oximetry monitoring in the operating theatre and in the recovery room. British Journal of Anaesthesia 68: 146-150. M011er JT, Johannessen NW, Espersen K et al (1993) Randomized evaluation of pulse oximetry in 20 802 patients. II. Perioperative events and postoperative complications. Anesthesiology 78: 445-453. Morgan M (1986) The Confidential Enquiry into Maternal Deaths. Anaesthesia 41: 689-691. Pedersen T (1994) Complications and death following anaesthesia--a prospective study with special reference to the influence of patient-, anaesthesia-, and surgery-related risk factors. Danish Medical Bulletin 41:319-331 (thesis). Pedersen T & Johansen SH (1989) Serious morbidity attributable to anaesthesia. Considerations for prevention. Anaesthesia 44: 504-508. Pedersen T, Eliasen K & Henriksen E (1990) A prospective study of mortality associated with anaesthesia and surgery: risk indicators of mortality in hospital. Acta Anaesthesiologica Scandinavica 34: 176-182. Report on the Confidential Enquiry into Maternal Deaths in the United Kingdom 1988-1990 (1994) London: HMSO. Tiret L, Desmonts JM, Hatton F et al (1986) Complications associated with anaesthesia: a prospective survey in France. Canadian Anaesthetists" Society Journal 33: 336-344. Utting JE, Gray TC & Shelley FC (1979) Human misadventure in anaesthesia. Canadian Anaesthetists' Society Journal 26: 472-478. Williamsen JA, Webb RK, Szehely S et al (1993) Difficult intubation. An analysis of 2000 incident reports. Anaesthesia and Intensive Care 21: 602-607. Zeitlin GL (1989) Possible decrease in mortality associated with anaesthesia: a comparison of two time periods in Massachusetts, USA. Anaesthesia 44: 432-433.