The incidence, detection and management of postoperative myocardial ischaemia
J. Noden and C. S. Reilly
Introduction
ment can be planned to minimise the risk of postoperative myocardial ischaemia.
Cardiac events such as myocardial infarction, unstable angina, congestive heart failure and arrhythmias, are the cause of one-third of all postoperative complications and are a factor in more than half of the deaths following surgery. A considerable amount of clinical research has been directed at the prediction of patients at risk of developing postoperative cardiac problems by pre-operative screening and peri-operative monitoring. Almost 20 years ago pre-operative clinical risk indices showed that patients with congestive cardiac failure, unstable angina or recent myocardial infarction were more likely to have postoperative cardiac e v e n t s / M o r e recently, pre-operarive detection of ischaemia by either ambulatory ECG monitoring or exercise stress testing and by more invasire methods such as dipyridamole-thallium scanning, or coronary angiography have been shown to be indicators of increased risk. Unfortunately, these tests may be impractical or unavailable for certain patients and may be too expensive for routine use. In the past few years, several studies have shown that postoperative myocardial ischaemia is a better predictor of cardiac morbidity in patients having major, non-cardiac surgery than any pre-operative factors examined) -5 In clinical practice, a major factor in reducing the incidence of postoperative ischaemia is the anaesthetist being aware of the possibility of pre-existing myocardial disease at the pre-operative visit. If the history and examination are geared towards identifying the presence of risk factors the anaesthetic and postoperative manage-
Incidence of postoperative myocardial ischaemia The incidence of postoperative myocardial ischaemia, as assessed using ambulatory ECG monitoring, varies between 30-63% (Table 1). A number of factors may influence the incidence including the patient's pre-operative state, anaesthesia and the type of surgery involved.
Pre-operativefactors Patients who have a history of coronary artery disease (CAD) or have a high risk of CAD, even if it is asymptomatic, are more likely to have postoperative ischaemia. It is interesting to note that patients with risk factors alone rather than proven CAD may be as likely to develop postoperative ischaemia as those with definite CAD? Definite CAD: • previous myocardial infarction
Table 1 Incidence of postoperative ischaemia and cardiac events
J. Noden MB ChB, FRCA, Registrar in Anaesthesia, C. S. Reilly MD, FRCA, Professor of Anaesthesia, University of Sheffield, Department of Surgical and Anaesthetic Sciences, Sheffield S 10 2JF, UK
Incidence of ischaemia
Incidenceof cardiacevents
Non-cardiac in at risk patients2 Non-cardiac in at risk patients3 Peripheral vascular surgery in patients with CAD21 Peripheral vascular surgery9 Lower extremity vascular surgery7 Major vascular surgery5 Peripheral vascular surgery in high risk patients8
42% 41%
13% 18%
63% 30% 40% 39.1%
37.5% 10% 8% 8.6%
55.5%
11.5%
CAD = CoronaryArtery Disease
Current Anaesthesia and Critical Care (1995) 6, 142-147
© Pe. . . . Professional Ltd 1995
Type of surgery
142
INCIDENCE,DETECTIONAND MANAGEMENTOF POSTOPERATIVEMYOCARDIALISCHAEMIA 143 • typical angina • atypical angina with ischaemic ECG response to exercise • scintographic evidence of a myocardial perfusion defect. Risk for CAD: • previous or current vascular surgery or • 2 or more cardiac risk factors: (+ male gender) age > 65 years hypertension current smoking raised serum cholesterol diabetes mellitus. If a patient has undergone successful coronary artery bypass surgery, the risk of postoperative ischaemia occurring after a subsequent non-cardiac operation is low and similar to that of a patient without significant CAD on angiography. The clinical significance of a brief period of ischaemia in healthy patients is unknown but repeated episodes may have a cummulative effect and eventually cause myocardial infarction. Obviously, these changes are more significant in patients with pre-existing heart disease.
Surgical factors Patients presenting for major vascular surgery are at increased risk of cardiac morbidity because of the increased probability of co-existing CAD. These patients have provided a major group for the study of postoperative ischaemia. Studies have demonstrated a high incidence of co-existent heart disease in this group of patients with 40-80% of patients presenting for aortic reconstruction having clinical evidence of CAD. There appears to be a greater risk of postoperative myocardial ischaemia following infrainguinal revascularisation than following aortic operations and adverse cardiac outcome occurs as frequently. 6 In patients having major surgery who are known to be at high risk of CAD, there is no difference in the incidence of postoperative ischaemia during the first 4 days between different types of procedure when comparing major vascular, abdomino-thoracic, orthopaedic, neurosurgical and head and neck surgery. However, ischaemia during the first week has a higher incidence following major vascular surgery. 1
receiving regional or general anaesthesia. 7 There was a similar incidence of major cardiac morbidity in both groups and, in both, this was significantly associated with the presence of postoperative ischaemia. However, epidural anaesthesia was associated with a lower incidence of re-operation for inadequate tissue perfusion and this may be an advantage for this population.
Postoperative factors (see Table 2) Myocardial ischaemia is dependent on the balance between myocardial oxygen supply and demand. There are a number of physiological changes in the postoperative period which have the potential to precipitate myocardial ischaemia in a susceptible patient by altering either or both of the oxygen supply and demand. The oxygen demand is increased by the cardiovascular changes associated with pain, altered catecholamine levels and the response to hypoxaemia and on the supply side the factors would include altered pulmonary function, body temperature changes, fluid balance changes and altered sleep patterns.
Heart rate Tachycardia produced by any of the above factors will result in an increase in myocardial oxygen demand and also in a decrease in the sub-endocardial perfusion time. This is particularly likely if there is concomitant increase in arterial pressure and LVEDP. The relationship between postoperative tachycardia and the incidence of myocardial ischaemia and infarction is not clear. A number of studies have shown a clear association between the incidence of tachycardia and the incidence of ischaemia. Also, however, a number of other studies have demonstrated no such association. Some case studies have shown the development of ischaemia following an acute increase in heart rate. Although some studies have found no significant difference in postoperative heart rate between patients who have and have not gone on to cardiac events, 5,8 others have demonstrated that in patients with postoperative ischaemia, overall peak heart rates were greater in those that had subsequent cardiac events. 9
Hypoxaemia This is common following major surgery. A reduction in functional residual capacity and the resultant shunting
Anaesthetic factors Table 2 Factorsassociatedwith postoperativeischaemia There have been a number of studies comparing regional and general anaesthesia with respect to postoperative morbidity and mortality. The overall impression is that there is little difference in early or late outcome between the two techniques. When looking at high risk patients undergoing peripheral vascular surgery a recent study demonstrated that there was no significant difference in incidence of postoperative ischaemia between patients
Factors associatedwith postoperativeischaemia
Association found (reference)
Preoperativeischaemia Treated hypertension Tachycardia Hypoxaemia Anaemia Hypothermia
9, 11, 22, 23 23 16 10, 16 8 11
144
CURRENT ANAESTHESIA AND CRITICAL CARE
leads to a reduction in arterial oxygen saturation. In addition, there may be acute decreases in saturation caused by the effect of opioids on ventilation and by disturbed sleep pattern. The principal causes of hypoxaemia change during the postoperative period. The effects of reductions in FRC and opioid administration are most likely in the first postoperative day followed by changes in sleep pattern, (particularly REM sleep), most likely on the third or fourth postoperative night. In patients with no previous evidence of CAD, episodic and persistant hypoxaemia following abdominal surgery have both been associated with tachycardia and the development of myocardial ischaemia. ~°
Anaemia A decrease in oxygen carrying capacity may lead to a direct decrease in coronary oxygen supply if adequate cardiovascular compensation cannot occur, i.e. in patients with limited functional coronary reserve. In patients at high risk of CAD undergoing infrainguinal arterial bypass procedures, anaemia on postoperative day 1 correlated significantly with postoperative ischaemia and morbid cardiac events?
Unintentional hypothermia Hypothermia is common postoperatively. Although used to advantage during some surgical procedures, it has many adverse physiological effects including cardiac arrhythmias, increased vascular resistance and a left shift in the Hb-oxygen saturation curve. Rewarming also places many demands on the cardiovascular system. In patients undergoing lower extremity vascular surgery, patients whose temperature was < 35°C immediately after surgery had a greater incidence of myocardial ischaemia than those who were above this temperature. This was independent of age. General anaesthesia resulted in more patients being hypothermic than if the operation was carried out under epidural but there was no difference in the incidence of ischaemia between the two groups. This effect was particularly noticeable in patients with no evidence of pre-operative ischaemia. This group had a four-fold increase in the incidence of ischaemia postoperatively when compared with pre-operatively? 1 In addition, more hypothermic patients had episodes of arterial hypoxaernia.
Study limitations In patients with or at high risk of CAD, pre-operative higher cardiac risk indices (e.g. Goldman) and evidence of pre-operative ischaemia are associated with postoperative morbid cardiac events. However, more recent work has suggested that the presence of postoperative ischaemia may be a better indicator of morbid cardiac events. In those studies that present the relevant information,
between 85% and 100% of patients who suffered a postoperative cardiac event had ECG evidence of preceding myocardial ischaemia postoperativelyY,8,9Unfortunately it is not possible to predict which of those with postoperative ischaemia are going to proceed to cardiac event with different studies finding an incidence between 18% and 50%. 7-9 Comparison between studies may be misleading as the frequency of occurrence of cardiac complications depends upon the method of detection and the precise definition of significant ischaemic events. Postoperative myocardial infarction is only one complication of ischaemia and its incidence is substantially underestimated because routine cardiac enzymes or 12-lead ECGs are not performed postoperatively on all patients undergoing surgery.
Detection of postoperative myocardial ischaemia An important part of the detection of postoperative myocardial ischaemia is to have an index of suspicion. The presence of pre-operative ischaemia correlates with the occurrence of postoperative ischaemia and Holter monitor evidence of pre-operative ischaemia increases the risk of postoperative adverse cardiac outcome. However, the use of other tests such as exercise stress testing, echocardiographic or radionucleide determination of ejection fraction, dipyridamole-thallium scanning, routine coronary angiography or monitoring methods including pulmonary artery catheterisation, multiple-lead ECG and transoesophageal echocardiography, does not necessarily identify those most likely to have cardiac complications. 6 Even ambulatory ECG has limitations because greater than 50% of adverse cardiac outcomes can occur in patients with no evidence of pre-operative ischaemia. The possible methods available for the detection of myocardial ischaemia postoperatively include clinical observation, 12-lead ECG, S-T trend analysis of the ECG and ambulatory ECG monitoring. In some circumstances transoesophageal echocardiography can be used. Detection of myocardial ischaemia by clinical observation is of very limited use. Postoperative myocardial infarction may be detected clinically by changes in arterial pressure or in rhythm but there is little association between myocardial ischaemia and clinical signs. Myocardial ischaemia most commonly presents as angina. Alternatively it may cause heart failure or arrhythmias. Unfortunately most of the recent studies have reported that postoperative myocardial ischaemia is clinically silent in 91-97% of cases. 2,3,5,12The reason for the high incidence of silent ischaemia is not clear but there are a number of factors present postoperatively that may mask it including surgical pain, residual anaesthetic agents, the administration of analgesics and fluid balance abnormalities. In addition, ambulatory patients who have episodes of silent ischaemia have been shown to have abnormal pain thresholds and pain perception?
INCIDENCE,DETECTIONAND MANAGEMENTOF POSTOPERATIVEMYOCARDIALISCHAEMIA 145 Postoperative patients also have a higher incidence of painless myocardial infarction (50-70%) than nonsurgical patients (20--40%). The use of serial 12-lead ECG recordings is also of fairly limited value in the detection of transient ischaemic episodes. The use of continuous ECG monitoring with S-T segment trend analysis has been introduced recently and has been shown to be of value in several studies. This may in future be a useful method for monitoring but it requires the patient to be in a high dependency unit to ensure regular observation of the trends. Ambulatory ECG monitoring has been shown to be a reliable method of detecting postoperative ischaemia. Monitors are now available which have an alarm which is activated by the presence of ischaemia and it is possible that such systems will be used increasingly in this setting.
Management of postoperative myocardial isehaemia The most important step in the management of postoperative ischaemia is to prevent its occurrence. The Perioperative Ischemia Research Group found that early postoperative ischaemia increases the odds of having any adverse cardiac outcome three-fold but, more importantly, there was a nine-fold increase in the odds of having further ischaemic events. They also showed postoperative ischaemia to be a more accurate predictor of these events than any pre-operative factors. If we can reduce the incidence of ischaemia postoperatively it should be possible to reduce the cardiac morbidity and mortality in these high risk patients.
Screening Prevention must begin pre-operatively with the aim of minimising the number of patients who have an unidentified risk of myocardial ischaemia. Time and resource restrictions often may make it impossible to be able to fully screen every patient pre-operatively. Simple clinical judgement will give an indication of who is more likely to be at risk: anybody with clinical CAD, patients who have undergone or are in need of peripheral vascular surgery or those who have more than one of the following risk factors; hypertension, hypercholesterolaemia, smoking and diabetes mellitus. The stress of major surgery on top of these conditions creates the potential for problems. More detailed screening tests have been advocated. Any at risk patient with severe angina should go through the process of non-invasive and, if necessary, invasive testing followed by, perhaps, even coronary angioplasty or bypass grafting before the proposed non-cardiac surgery is undertaken. This has been shown to reduce morbidity and mortality in patients with CAD requiring non-urgent, non-cardiac surgery. 13It is important that any correctable factors such as anaemia, hypertension or coexistent
respiratory disease are corrected and the drug therapy altered to achieve optimal pre-operative conditions. There must be effective pre-operative communication between surgeon, physician/cardiologist and anaesthetist. Unless emergency surgery is required if current medical treatment is not fully controlling symptoms, adequate time must be allowed for any changes to be implemented and results assessed.
The intra-operative period Induction and reversal of anaesthesia stress the cardiovascular system. The routine use of intra-operative haemodynamic monitoring allows the anaesthetist to maintain cardiovascular variables within more strict predetermined limits and thus avoid large fluctuations in myocardial oxygen supply and demand. This, plus the careful choice of anaesthetic drugs and technique, has resulted in the incidence of intra-operative ischaemia being no higher than that pre-operatively and certainly less than that seen postoperatively?,9 The use of drugs with a low potential for causing changes in haemodynamics and a low risk of histamine release, such as etomidate, vecuronium and fentanyl is advisable. There is little evidence to demonstrate a significant difference in development of ischaemia or in outcome between the use of general anaesthesia and local analgesic techniques. Prevention of hypothermia must be commenced intraoperatively. This is a very simple way of reducing one of the risk factors of myocardial ischaemia and really ought to be considered in any major procedure if the patient is at risk. For any operation that is going to involve considerable blood loss, the anaesthetist should insist on the careful measurement of the loss and have a low threshold for intervention with transfusion. Nelson found that a haematocrit below 28% was associated with an increased incidence of ischaemia. 8 Careful monitoring of the patient's haemodynamic stability peri-operatively is essential. In particular, episodes of hypo- or hypertension and/or tachycardia should be treated immediately.
Postoperatively Ideally, any patient who is deemed at risk of cardiac complications should be closely supervised postoperatively and should be cared for on an intensive care or high dependency unit. In these settings, invasive haemodynamic monitoring can be continuous so that any deviations from acceptable values are picked up and acted upon. These facilities may not be present in all hospitals but it is the responsibility of the anaesthetist to organise the postoperative care of their at risk patient. In doing this, attention must be paid to the physiological and pharmacological factors which might influence the incidence of ischaemia including attention to the haemodynamics, oxygenation, fluid balance and analgesia of the patient.
146
CURRENTANAESTHESIAAND CRITICALCARE
Haemodynamics Hypertension, hypotension and tachycardia increase the risk of developing ischaemia and should be avoided. Any patient who is at risk should be closely monitored and there should be guidelines for treatment of any values that fall outside predetermined levels. Nursing staff should be given instructions to inform medical staff as soon as deviations occur so that interventions can be at an early stage. A patient should not be left with a tachycardia or a high blood pressure for any length of time in the hope that it 'may settle'. As described previously the combination of hypertension and tachycardia is undesirable. The most likely causes of such haemodynamic disturbances are pain, hypoxaemia and hypercarbia. It is essential that any patient with hypertension and tachycardia is assessed promptly to elicit the cause and that appropriate treatment is started immediately. Studies which have used invasive monitoring in the immediate postoperative period with a rigid protocol on intervention when there were any changes in blood pressure or heart rate have shown a clear benefit in outcome? 4
to reduce the severity of ischaemic episodes in patients who have undergone myocardial revascularisation. 18 High dose opioids can be used safely in patients with poor cardiac function and are good at controlling intraoperative haemodynamics. Few studies have examined the effect which different analgesic regimens may have on postoperative ischaemia following non-cardiac surgery. However, it is obvious that pain will lead to hypertension and tachycardia and that adequate analgesia will decrease sympathetic stimulation and myocardial oxygen consumption. It is also clear that over sedation and respiratory depression produced by inadvertent overdosage of opioids is likely to produce hypoxaemia and therefore increase the risk of ischaemia. Surgery under epidural anaesthesia does not cause the rise in catecholamines that occurs during general anaesthesia. A study looking at ischaemia following vascular surgery showed no difference in the incidence between epidural or general anaesthesia but analgesia was not maintained postoperatively with epidural local anaesthetic. ~9Further research is required on the influence of analgesic drugs and modes of delivery on the incidence of ischaemia? °
Hypoxaemia The relationship between episodic hypoxaemia and postoperative myocardial infarction is unclear but studies have shown that hypoxaemia and tachycardia are involved in the development of postoperative myocardial ischaemia in patients both with and without cardiovascular disease. Myocardial ischaemia is more likely to occur if an episode of hypoxaemia is prolonged and severe. 15,16 Myocardial ischaemia and infarction occur most commonly on the third postoperative day which is when arterial oxygen saturation may still be low and supplementary oxygen is not used routinely. One reason for this is the return of REM sleep that has been suppressed on the first and second postoperative nights, particularly by opioids. REM sleep is associated particularly with sleep apnoea and episodic desaturation. Consequently, in patients following major surgery oxygen should be given for 24 h, then overnight for the next 2 nights to reduce the hypoxaemic episodes. However, silent ischaemia in patients who have chronic stable angina shows a diurnal distribution with a peak in the morning that corresponds to a similar pattern of heart rate, plasma catecholamine levels and indices of coagulation. 17A similar distribution has been reported for sudden cardiac death and acute myocardial infarction. This diurnal variation has also been shown to occur postoperatively in patients who have had hip arthroplasty under regional anaesthesia and in patients having general anaesthesia for non-cardiac surgery? z The pattern and requirement for postoperative oxygen therapy will be determined by the patient's pre-operative condition and the nature of the surgery undertaken.
Analgesia The use of intensive analgesic regimes has been shown
Drug therapy It is important that the patient's cardiovascular drug therapy is continued throughout the peri-operative period or that therapy is omitted for the shortest time period possible. This may be difficult if a surgical procedure limits administration by the oral route. However, Betaadrenergic blockers such as propranolol or labetolol can be given by IV infusion. Nitrates may be given by infusion or transdeIa'nally if required. Acute myocardial infarction is usually due to rupture of an atheromatous plaque and there is no evidence that preventing ischaemic episodes has any effect on this process. However, if increasing frequency of myocardial ischaemia is taken as a sign of plaque instability, antithrombotic agents or smooth muscle relaxants may be of use. Intense procoagulant activity and sympathetic stimulation postoperatively may lead to coronary vasospasm, thrombosis or plaque rupture. 12 Heparin, either prophylactically or as part of surgical management, may have a beneficial effect on myocardial ischaemia as it has been shown to decrease frequency of anginal attacks and silent ischaemic episodes in patients with unstable angina. More recently it has been thought that postoperative myocardial complications may be due to long-duration subendocardial ischaemia with many of the infarctions being non-Q-wave. 5 To reduce the number of ischaemic episodes perhaps we ought to be using anti-anginal drugs more widely as these are also effective in treating silent ischaemia. Beta- adrenergic blockers alter the circadian distribution of myocardial ischaemia by removing the morning peak and trials are underway to see if they alter prognosis in medical patients with stable angina and ambulatory ischaemia. ~7 Similar trials may be useful in postoperative patients.
INCIDENCE, DETECTION AND MANAGEMENT OF POSTOPERATIVE MYOCARDIAL ISCHAEMIA
The use of on-line ST-segment monitoring and analysis will allow early recognition and prompt intervention of ischaemia especially when there is no preoperative history of angina. References 1. Goldman L, Caldera D L, Nussbaum S R et al. Multifactorial index of cardiac risk in noncardiac surgical procedures. N Engl J Med 1977; 297: 845-850. 2. Mangano D T, Hollenberg M, Fegert Get al. Perioperative myocardial ischemia in patients undergoing noncardiac surgeryl:incidence and severity during the 4 day perioperative period. JACC 1991; 17: 843-850. 3. Mangano D T, Browner W S, Hollenberg M e t al. Association of perioperative myocardial ischaemia with cardiac morbidity and mortality in men undergoing noncardiac surgery. N Engl J Med 1990; 323: 1781-1788. 4. Fleisher L A, Rosenbaum S H, Nelson A H et al. The predictive value of preoperative silent ischaemia for postoperative ischaemic cardiac events in vascular and nonvascular surgery patients. Am Heart J 1991; 122: 980-986. 5. Landesberg G, Luria M H, Cotev S e t al. Importance of longduration postoperative ST-segment depression in cardiac morbidity after vascular surgery. Lancet 1993; 341 : 715-719. 6. Krupski W C, Layug E L, Reilly L M e t al. Comparison of cardiac morbidity between aortic and infralnguinal operations. J Vasc Surg 1992; 15: 354-365. 7. Christopherson R, Beattie C, Frank S M e t al. Perioperative morbidity in patients randomized to epidural or general anesthesia for lower extremity vascular surgery. Anesthesiology 1993; 79: 422-434. 8. Nelson A H, Fleisher L A, Rosenbanm S H. Relationship between postoperative anemia and cardiac morbidity in high-risk vascular patients in the intensive care unit. Crit Care Med 1993; 21 : 860-866. 9. Raby K E, Barry J, Creager M A, Cook F, Weisberg M C, Goldman L. Detection and significance of intraoperative and postoperative myocardial ischemia in peripheral vascular surgery. JAMA 1992; 268: 222-227. 10. Rosenberg J, Rasmussen V, von Jessen F, Ullstad T. Late
147
postoperative episodic and constant hypoxaemia and associated ECG abnormalities. Br J Anaes 1990; 65: 684-69l. 11. Frank S M, Beattie C, Christopherson R E et al. Unintentional hypothermia is associated with postoperative myocardial ischemia. Anesthesiology 1993; 78: 468--476. 12. Marsch S C U, Schaefer H G, Skarvan K, Castelli I, Scheidegger D. Perioperative myocardial ischemia in patients undergoing elective hip arthroplasty during lumbar regional anesthesia. Anesthesiology 1992; 76: 518-527. 13. Mahar L J, Steen P A, Tinker J H et al. Perioperative myocardial infarction in patients with coronm2¢ artery disease with and without aorta-coronary artery bypass grafts. J Thor Cardiovasc Surg 1978; 76: 533-537. 14. Rao T L, Jacobs K H, EI-Etr A A. Reinfarction following anaesthesia in patients with myocardial infarction, Anesthesiology 1983; 59: 499-505. 15. Gill N P, Wright B, Reilly C S. Relationship between hypoxaemic and cardiac ischaemic events in the perioperative period. Br J Anaes 1992; 68: 471-473. 16. Reeder M K, Muir A D, Frex P e t al. Postoperative myocardial ischaemia: temporal association with nocturnal hypoxaemia. Br J Anaes 1991; 67: 626-631. 17. Fox K M, Mulcahy D A. Therapeutic rationale for the management of silent ischemia. Circulation I990; 82(Suppl. 2): 155-160. 18. Mangano D T, Siliciano D, Hollenberg M e t al. Postoperative myocardial ischernia:therapeutic trials using intensive analgesia following surgery. Anesthesiology 1992; 76: 342-353. 19. Yeager M P, Glass D D, NeffR K et al. Epidural anaesthesia and analgesia in high risk surgical patients. Anesthesiology 1987; 66: 729-736. 20. Reilly C S. Regional analgesia and myocardial ischaemia. Br J Anaes 1993; 71: 467-468. 21. Ouyang P O, Gerstenblith G, Furman W R et al. Frequency and significance of early postoperative silent myocardial ischaemia in patients having peripheral vascular surgery. Am J Cardiol 1989; 64: 1113-1116. 22. Raby K E, Goldman L, Creager M A et al. Correlation between preoperative ischemia and major cardiac events after peripheral vascular surgery. N Engl J Med 1989; 321: 1296-1300. 23. McHugh P, Gill N P, Wyld R, Nimmo W S, Reilly C S. Continuous ambulatory ECG monitoring in the perioperative period:relationship of preoperative status and outcome. Br J Anaes 1991; 66: 285-291.