Evaluation of Anesthetic Risk
Vijaya Gottumukkala
Dam-Thuy Truong
Angela T. Truong
CASE SUMMARY
A 50-year-old, morbidly obese, male patient with renal insufficiency, a history of transient blindness in the right eye, and chronic stable angina class II was scheduled to undergo an elective lumbar vertebrectomy (L1-4) and posterior spine stabilization. As part of the preoperative workup, the patient had a stress thallium study, which was positive for inducible ischemia in the area of the left anterior descending (LAD) artery. After an angiogram showed 80% stenosis in the LAD artery, he underwent an uneventful coronary angioplasty with placement of a bare metal stent in the LAD. Thereafter, he began to experience symptoms of spinal cord compression, and surgery was urgently performed 6 weeks after the angioplasty. Aspirin was continued perioperatively but clopidogrel was discontinued. The surgery lasted 8 hours, and four units of packed red blood cells were transfused intraoperatively. The patient experienced brief periods of hypotension during the procedure, which responded to fluid therapy and intermittent boluses of ephedrine. He was transferred to an intensive care unit (ICU) for postoperative care after the procedure. His trachea was successfully extubated the next morning. He complained of reduced vision in his left eye along with weakness and reduced sensation in his upper right arm on the first postoperative day, and was diagnosed with posterior ischemic optic neuropathy and brachial plexopathy. On the second postoperative day, he suffered an acute anterior wall myocardial infarction (MI) caused by stent thrombosis, along with transient blindness in the right eye, which quickly resolved. The patient’s hospital course was further complicated by a chest infection that necessitated antibiotic treatment and noninvasive ventilation. The patient was discharged on postoperative day 45 with some residual upper extremity weakness and improving visual acuity. Six months after the surgery, the visual acuity in the left eye had improved to 20/25, with a persistent left temporal field deficit.
The questions we, as anesthesiologists, should ask ourselves are as follows: Could any of these complications have been predicted? Was the preoperative workup and preparation appropriate? Could the perioperative complications have been avoided?
How Is Anesthesia Risk Quantified and Predicted?
Risk is defined as the possibility of suffering harm or loss (also see Chapters 1 and 3). Why study anesthesia risk? First, because “do no harm”—primum non nocere—is a fundamental doctrine in the practice of medicine. Second, as part of a medical specialty that has championed patient safety, our quest as anesthesiologists is to reduce anesthetic-related mortality and accurately predict the likelihood of complications in the perioperative period for a patient undergoing a particular surgical procedure. Finally, through active interventions, our intent is to reduce the incidence of complications and improve outcome by positively modifying the impact of the inherent risk factors.
Since anesthesia is rarely therapeutic and always administered to facilitate a diagnostic or therapeutic intervention, the evaluation of anesthesia risk is, in fact, the assessment of perioperative risk. Therefore, any study of risk during the perioperative period involves understanding the interactions between the patient’s comorbid conditions and functional status, the surgical technique, surgeon’s skill, complexity of the surgical procedure, and, finally, the anesthetic and postoperative management.
It is challenging to accurately quantify and reliably predict anesthesia risk through a review of the available literature for the reasons outlined in Table 2.1.
Given these challenges, a systematic and standardized approach to studying risk is much needed. Our goal should be to accurately predict and modify the risk factors, if possible, and to manage patients appropriately with the state-of-the art technology and best evidencebased practices to achieve optimal outcomes for our patients.
TABLE 2.1 Reasons for Difficulties in Reliable Prediction and Quantification of Anesthetic Risk | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
|
What Is the Historical Perspective on Anesthesia Risk?
▪ POSTOPERATIVE MORBIDITY AND MORTALITY
General Anesthesia
The first description of anesthetic-related deaths was apparently by John Snow in a treatise titled On Chloroform and Other Anesthetics.1 Early studies on anesthesia-related postoperative mortality reported an incidence of 1 in 1,500 surgical procedures.2,3,4,5 These studies concluded that errors in surgical diagnosis, judgment, and/or technique were the most common causes for immediate postoperative deaths. Other factors in decreasing order of frequency were the patient’s medical condition and the anesthetic technique.
Marx et al.6 and Dripps et al.7 independently reported that regional anesthesia was associated with a lower rate of mortality than general anesthetic techniques. In the later report by Dripps et al.,7 deaths deemed related to spinal anesthesia occurred in 1 of 1,560 procedures, and deaths possibly related to spinal anesthesia occurred in 1 of 780 procedures. Corresponding incidences for general anesthesia were 1 of 536 and 1 of 259, respectively.7
Farrow et al.8,9 from Wales and Harrison10 from South Africa independently evaluated mortality associated with surgery in the 1970s and 1980s. Farrow et al. reported that anesthesia-related mortality was highest in patients older than 65 years and anesthesia contributed to 2.2% of perioperative mortality. Harrison10 noted that anesthesia-related deaths had significantly decreased in the later stages of their study as compared with the data gathered during the initial phases. They attributed this positive change to improvements in routine monitoring, better supervision of trainees, a decrease in caseload per anesthesiologist, and the introduction of recovery rooms and ICUs during their study period. Holland11 from New South Wales, Australia, between 1960 and 1985 reported on deaths occurring within 24 hours of an anesthetic. The mortality rate decreased from 1 in 5,500 in the 1960s to 1 in 10,250 in the 1970s and to 1 in 26,000 in 1984. The three most common causes of death in their report were inadequate preparation of the patient and inadequate intraoperative and postoperative crisis management. The direct result of this study in Australian anesthesia practice was the phasing out of the resident medical officer as an anesthesia work force member.
Studies were also conducted during the same period in France by Tiret et al.12 and a decade later in Finland by Tikkanen and Hovi-Viander.13 The French study reported that respiratory depression in the postoperative period was the predominant cause of death and coma (1 in 13,207) and that major complications occurred more frequently in older patients, those with multiple comorbid conditions, and patients undergoing emergency surgical procedures. Mortality associated with anesthesia and surgery was studied in Finland in 1986 using a retrospective method, and the results were compared with those of a similar study performed in 1975. Surgery was the main contributing factor in the deaths of 22 patients (frequency 0.68/10,000 procedures) and anesthesia in the deaths of 5 patients (frequency 0.15/10,000 procedures). The role of surgery in 1986 had decreased to approximately one third and the role of anesthesia to less than one tenth as the main causes of death associated with anesthesia and surgery compared with the year 1975; in fact, 95.3% of all the patients died mainly because of coexisting medical or surgical conditions. The significant changes in anesthetic practice between the two study periods were a significant increase in specialist anesthesiologists and better recovery room and intensive care facilities.
Earlier studies of anesthetic-related mortality in the United Kingdom by Lunn et al.14,15,16 led to the development of the National Confidential Enquiry into Perioperative Deaths (NCEPOD) in 1987.17 All deaths that occurred within 30 days of surgery were included in the inquiry. The study categorized deaths as related to either anesthesia or surgery. There were 4,034 deaths over a 12-month period in 485,850 surgeries, leading to a crude mortality rate of 0.7% to 0.8%. Surgery contributed directly or partially to 30% of all deaths. Anesthesia was considered the sole cause of death in only 3 patients, leading to a mortality rate of 1 in 185,000, and was contributory in 410 deaths, for a rate of 7 in 10,000 cases. A striking finding of the report was that 20% of the deaths were avoidable. Avoidable factors for both anesthesiologists and surgeons were a lack of adequate trainee supervision (approximately 50% of deaths in one of the regions were those of patients who had no consultant contributing to their clinical care); failure
to act appropriately with existing knowledge (rather than lack of knowledge); equipment malfunction; and fatigue.17 At the same time, a Canadian study reported that no deaths could be directly attributable to anesthesia.18 Four factor groups (patient, surgical, anesthesia, and “other”) were assessed by logistic regression analysis to ascertain which variables were predictive of death within a week of surgery. Advanced age, male gender, compromised physical status, major surgery, emergency procedure, procedures performed from 1975 to 1979, intraoperative complications, narcotic techniques, and having one or two anesthetic drugs administered were associated with a higher mortality rate, whereas duration of anesthesia, experience of the anesthesiologist, and inhalation techniques did not affect the mortality rate. The authors concluded that the patient and surgical risk factors were much more important in predicting 7-day mortality than the anesthesia factors studied.18
to act appropriately with existing knowledge (rather than lack of knowledge); equipment malfunction; and fatigue.17 At the same time, a Canadian study reported that no deaths could be directly attributable to anesthesia.18 Four factor groups (patient, surgical, anesthesia, and “other”) were assessed by logistic regression analysis to ascertain which variables were predictive of death within a week of surgery. Advanced age, male gender, compromised physical status, major surgery, emergency procedure, procedures performed from 1975 to 1979, intraoperative complications, narcotic techniques, and having one or two anesthetic drugs administered were associated with a higher mortality rate, whereas duration of anesthesia, experience of the anesthesiologist, and inhalation techniques did not affect the mortality rate. The authors concluded that the patient and surgical risk factors were much more important in predicting 7-day mortality than the anesthesia factors studied.18
The same authors19 reported the results of a followup study in 27,184 anesthetic procedures performed on inpatients in four major teaching hospitals in Canada. Again, no deaths were directly attributed to anesthesia. Data on these patients were collected, and the outcomes determined for the intraoperative, immediate postanesthetic, and postoperative time periods. Logistic regression was used to control for differences in patient populations across all four hospitals. There were large variations (two-to five-fold after case-mix adjustment) in minor outcomes across the four hospitals. The rates of major events and deaths were similar in three hospitals. One of the hospitals had a significantly lower mortality rate but a significantly higher rate of all major events (cardiac arrest, MI, and stroke). Possible reasons for these differences include lack of compliance in recording events, inadequate case-mix adjustment, differences in interpretation of the variables (despite guidelines), and institutional differences in monitoring, charting, and observation protocols. The authors concluded that measuring the quality of care in anesthesia by comparing major outcomes is unsatisfactory because the contribution of anesthesia to perioperative outcomes is uncertain, and variations may be explained by institutional differences that are beyond the control of the anesthesiologist.
Warner et al.20 studied major morbidity and mortality occurring within 1 month of ambulatory surgery in 38,598 patients and reported no deaths directly attributable to anesthesia. Fleisher et al.21 performed a claims analysis of a nationally representative sample of Medicare beneficiaries for the years 1994 to 1999. Patients undergoing 16 different surgical procedures under anesthesia in an outpatient hospital, ambulatory surgery center, and a physician office were studied. The 7-day mortality rate was 50 per 100,000 in the outpatient hospital, 35 per 100,000 in the office setting, and 25 per 100,000 in the ambulatory surgery setting.
Wolters et al.22 prospectively studied 6,304 consecutive patients admitted to surgery for perioperative complications in Germany. The objective of their study was to correlate variables recorded perioperatively with morbidity and mortality in an attempt to assess the predictive value of the variables for the outcome of individual patients. Data collected were American Society of Anesthesiologists (ASA) physical status, emergency or elective surgery, presence of medical comorbid conditions, type of anesthesia, type of surgery, and the operating time. The surgeries were classified by the Hoehn system (commonly used in Germany) into minor, moderate, or major surgeries. Logistic regression analysis was used to study the correlation between the preoperative risk factors and postoperative complications. Within the study group of 6,304 patients, 140 died postoperatively, and 1,432 patients developed complications that they survived. The variables that had the most influence on the risk for postoperative complications were ASA class IV (odd ratio [OR], 4.2), followed by ASA class III (OR, 2.2), and severity of surgery (OR, 1.86). The model described by Wolters et al.22 is 96% specific (able to predict uncomplicated course correctly predicted in 96% of cases) but has a very low sensitivity of only 16% (ability to correctly predict complications), giving a positive predictive value of 57% and a negative predictive value of 80%. The authors concluded that despite studying a large number of variables, they were unable to predict the risk of complications for individual patients with any degree of accuracy using statistical methods. They also implied that basing the surgical care of a patient on the probability of complications predicted by a statistical model is not any more accurate or reliable than qualified clinical judgment.
The results of a national confidential inquiry into perioperative deaths conducted in France were reported in 2004.23 In this report, an expert committee anonymously analyzed the patients’ charts to determine the cause of death and its relation to anesthesia. The annual rates of deaths that were totally or partially related to anesthesia were 7 (95% confidence intervals [CI]; 95% CI, 2-12) and 47 (95% CI, 31-63) per million, respectively. These mortality rates increased with patient comorbidity from 4 per million in patients of ASA physical status class I to 554 per million for those in ASA class IV. Similarly, the rates increased with age from 7 per million in patients younger than 45 years to 32 per million in older patients. Accidents were of respiratory (38%), cardiac (31%), ischemic (25%), and vascular origin (30%). The main surgical procedures that contributed to the mortality were orthopedic (50%: hip fracture, hemorrhagic surgery) and colorectal (24%: occlusion, peritonitis).
Institut National de la Santé et de la Recherche Médicale reported data between 1978 and 1982 on annual complications associated with anesthesia and found that 76 and 263 per million, respectively, were totally or partially related to anesthesia. The two aforementioned studies indicate that the anesthesia-related mortality rate has decreased 10-fold during the subsequent study periods, whereas the number of anesthetic procedures doubled. In addition, the number of procedures involving elderly patients and patients with poor physical status was four times higher. The authors attributed these improved results to enhanced safety and adherence to practice guidelines published after 1982. They hoped that the rate of 1 per 145,000 (at the time of publication) would serve as a basis for systematic analysis of accidents and review of future trends.
Kawashima et al.24 reported the results of anesthesiarelated mortality and morbidity from a confidential inquiry over a 5-year period in 2,363,038 patients from Japan in 2003. Data were analyzed for the incidence of critical events during anesthesia and surgery and the outcomes of the events within 7 postoperative days. The average annual incidences of cardiac arrests during surgery due to all etiologies and totally attributable to anesthesia were 7.12 (95% CI, 6.30, 7.94) and 1.00 (95% CI, 0.88, 1.12) per 10,000 cases, respectively. The average annual mortality within 7 postoperative days due to all etiologies and that totally attributable to anesthesia were 7.18 (95% CI, 6.22, 8.13) and 0.21 (95% CI, 0.15, 0.27) per 10,000 cases, respectively. Two principal causes of cardiac arrest during anesthesia and surgery due to all etiologies were massive hemorrhage (31.9%) and surgery-related factors (30.2%); those totally attributable to anesthesia were drug overdose or selection error (15.3%) and serious dysrhythmias (13.9%). This report indicates that preventable human errors caused 53.2% of cardiac arrests, and 22.2% of those deaths in the operating room were totally attributable to anesthesia. The authors concluded that the rates of cardiac arrest and death during anesthesia and surgery due to all etiologies, as well as those totally attributable to anesthesia, were comparable to those of other developed countries.
The National Surgical Quality Improvement Program (NSQIP) is the first national, validated, outcome-based, risk-adjusted, and peer-controlled program for measurement and enhancement of the quality of surgical care in the Veterans Administration (VA) System. Khuri et al.25 reported the results of the first prospective multicenter study that identified the predictors of 30-day and longterm survival after major surgery, taking into account the compendium of preoperative, intraoperative, and postoperative variables that define an episode of surgical care from the NSQIP database. This study showed that events in the postoperative period are more important than preoperative patient risk factors in determining the survival after major surgery in the VA. After correcting for the confounding variables collected in the NSQIP, the occurrence of a complication within the first 30 days postoperatively, independent of the patient’s preoperative risk, reduced median patient survival by 69% in the total patient study group. This study did not focus on the role of anesthetic management in the survival rate of patients after major surgery.
The number of deaths identified each year by the NCEPOD changed little between 1989 and 2003. Successive reports showed that most deaths occur in older patients who undergo major surgery and who have severe coexisting disease.26 A recent analysis identified higher mortality rates in a UK hospital when compared with a similar institution in the United States.27 Pearse et al.28 extracted data from the Intensive Care National Audit and Research Center (ICNARC) database and the Clinical Accountability, Service Planning and Evaluation Health Care Knowledge Systems database on inpatient general surgical procedures and ICU admissions in 94 National Health Service hospitals in the United Kingdom between January 1999 and October 2004. There were a total of 4,117,727 surgical procedures, of which a high-risk population of 513,924 patients was identified. These high-risk patients accounted for 83.8% of deaths but for only 12.5% of procedures. Despite the high mortality rates, <15% of these patients were admitted to the ICU (see Fig. 2.1). The authors concluded that the incidence of postoperative deaths in the United Kingdom has changed little in recent years, and that most deaths occur in older patients with coexisting medical disease who undergo major surgery. They recommended that higher ICU resources be provided for these high-risk patients.
 FIGURE 2.1 Mortality rates for general surgical patients identified from the * Clinical Accountability, Service Planning and Evaluation Health Care Knowledge Systems (CHKS Ltd.) and ICNARC databases. CHKS database: standard, all patients admitted to hospital for a general surgical procedure with an overall mortality rate of <5%; high risk, subpopulation of patients undergoing a procedure with an overall mortality rate of 5% or more. ICNARC database: intensive care unit (ICU), general surgical patients admitted directly to the ICU following surgery; ward to ICU, patients admitted to the ICU following initial postoperative care on a standard ward. (From Pearse RM, Harrison DA, James P, et al. Identification and characterisation of the high-risk surgical population in the United Kingdom. Crit Care. 2006;10:R81.) |
Regional Anesthesia
Auroy et al.29 reported the results of a 10-month prospective survey of 8,150 French anesthesiologists on major complications related to regional anesthesia based on (a) voluntary reporting during the study period using a telephone hotline service available 24 hours a day and managed by three experts; and, (b) voluntary reporting of the number and type of regional anesthesia procedures performed using pocket booklets. A total of 487 participants reported 56 major complications in 158,083 regional
anesthesia procedures (3.5 per 10,000). Four deaths were reported: three after spinal anesthesia and one after a posterior lumbar plexus block. Cardiac arrests occurred only after spinal anesthesia (n = 10; 2.7 per 10,000) and posterior lumbar plexus block (n = 1; 80 per 10,000). Systemic local anesthetic toxicity in all cases consisted of seizures only, without any cardiac events. Lidocaine spinal anesthesia associated with more neurologic complications than bupivacaine (14.4 per 10,000 vs. 2.2 per 10,000). Most neurologic complications were transient in nature. Among the 12 complications that occurred after peripheral nerve blocks, 9 were in patients in whom a nerve stimulator had been used.
anesthesia procedures (3.5 per 10,000). Four deaths were reported: three after spinal anesthesia and one after a posterior lumbar plexus block. Cardiac arrests occurred only after spinal anesthesia (n = 10; 2.7 per 10,000) and posterior lumbar plexus block (n = 1; 80 per 10,000). Systemic local anesthetic toxicity in all cases consisted of seizures only, without any cardiac events. Lidocaine spinal anesthesia associated with more neurologic complications than bupivacaine (14.4 per 10,000 vs. 2.2 per 10,000). Most neurologic complications were transient in nature. Among the 12 complications that occurred after peripheral nerve blocks, 9 were in patients in whom a nerve stimulator had been used.
Irita et al.30 investigated critical incidents associated with regional anesthesia using data from annual surveys of the membership of the Japanese Society of Anesthesiologists between 1999 and 2002. The total numbers of anesthetics available for analysis were 3,855,384, of which spinal anesthesia, combined spinal-epidural anesthesia, and epidural anesthesia contributed to 409,338, 146,282, and 69,001 reviews, respectively. A total of 628 critical incidents were reported in patients receiving regional anesthesia, including 108 cardiac arrests and 45 subsequent deaths. The causes of critical incidents were classified as totally attributable to anesthetic management or due mainly to intraoperative pathologic events, preoperative complications, or surgical management. Intraoperative pathologic events consisted of coronary ischemia, including coronary vasospasm (not suspected preoperatively), dysrhythmias (including severe bradycardia), pulmonary thromboembolism, and other conditions. Mortality was defined as death within 7 postoperative days. The incidences of cardiac arrest and mortality due to all causes were 1.69 per 10,000 and 0.76 per 10,000 with spinal anesthesia, 1.78 per 10,000 and 0.68 per 10,000 with combined spinal-epidural anesthesia, and 1.88 per 10,000 and 0.58 per 10,000 with epidural anesthesia, respectively. The incidences of cardiac arrest and mortality due to anesthetic management were 0.54 per 10,000 and 0.02 per 10,000 with spinal anesthesia, 0.55 per 10,000 and 0.00 per 10,000 with combined spinalepidural anesthesia, and 0.72 per 10,000 and 0.14 per 10,000 with epidural anesthesia, respectively. These values did not significantly differ between the three regional anesthesia groups. Intraoperative pathologic events and anesthetic management were responsible for 43% and 33% of cardiac arrests, respectively. Among the causes of anesthetic management-induced critical incidents, inadvertent high spinal anesthesia was the leading cause of cardiac arrest in both spinal and combined spinal-epidural anesthesia groups. Ninety percent of cardiac arrests occurred in patients with ASA physical status I and II, 88% in patients below 65 years of age, 45% in patients undergoing hip or lower extremity surgery, and 25% in patients undergoing cesarean section. The authors concluded that the incidence of cardiac arrest and mortality associated with central neuraxial anesthesia in Japan was similar to that of other developed countries. In patients with good ASA physical status, critical incidents occurred more often under regional anesthesia than under general anesthesia.
Summing Up
Lagasse31 reported that the overall anesthesia-related mortality rate in a suburban university hospital network in New York has remained stable at one death per 13,000 procedures over the last decade and that mortality increased with a higher score in the ASA classification system. This finding is in contrast to the NCEPOD17 and other reports19,20,23 that have shown much lower rates of anesthetic-related mortality in recent years. The reason for the discrepancy in wide variation in the reported incidence of both postoperative complications and anesthesia-related morbidity/mortality appears to be due to differences in the operational definition, measurement, and reporting of measures of anesthesia risk. Risk-adjusted comparative measures are very important in deriving any meaningful information in such a complex task. In spite of a wealth of literature on perioperative mortality, the challenge still remains to accurately predict the risk of major complications in a given patient for a particular procedure, as those two are the most important factors that determine outcome (see Fig. 2.2).
Perioperative Cardiac Risk
Cardiac and cerebrovascular diseases are the leading causes of death worldwide and represent a critical health problem.32 It is estimated that 25 million patients
anesthetized annually in the United States are at risk for or have clinically significant cardiovascular disease. More than 50% of the deaths after surgery are related to cardiac events.33 In patients who have ischemic heart disease, the incidence of perioperative MI following noncardiac surgery is 5.6%.34 The incidence of cardiac adverse events in high-risk patients undergoing major vascular surgery is higher at 10% to 18%.35 Raby et al.36 showed that the odds ratio (OR) for adverse cardiac events increased significantly to 2.7 in patients with preoperative ischemia and to 16 in those with postoperative ischemia. The perioperative morbidity and mortality as a result of this complex disease process can be reduced by a comprehensive clinical evaluation, use of reliable cardiac risk indexes, and appropriate risk-modifying measures.
anesthetized annually in the United States are at risk for or have clinically significant cardiovascular disease. More than 50% of the deaths after surgery are related to cardiac events.33 In patients who have ischemic heart disease, the incidence of perioperative MI following noncardiac surgery is 5.6%.34 The incidence of cardiac adverse events in high-risk patients undergoing major vascular surgery is higher at 10% to 18%.35 Raby et al.36 showed that the odds ratio (OR) for adverse cardiac events increased significantly to 2.7 in patients with preoperative ischemia and to 16 in those with postoperative ischemia. The perioperative morbidity and mortality as a result of this complex disease process can be reduced by a comprehensive clinical evaluation, use of reliable cardiac risk indexes, and appropriate risk-modifying measures.
 FIGURE 2.2 Mortality rate per 1,000 anesthetics reported by different studies. |
Are Risk Indexes Reliable Predictors of Outcome?
One of the earliest and most widely used clinical indexes for risk stratification is the ASA physical status classification.7 However, the ASA physical status was never intended to be an outcome predictor. Goldman et al.37 have developed a multifactorial index of cardiac risk for noncardiac surgical procedures. This method generated much interest and has led to numerous other indexes evaluating perioperative risk for both cardiac and noncardiac procedures.38,39,40,41,42,43 Most of these indexes are multifactorial and developed from observational studies. They are not widely used in everyday clinical practice because of their complexity and inaccuracy in predicting outcome for individual patients. Anesthesiologists and other perioperative health care providers commonly use ASA and New York Heart Association classification for risk stratification. However, these two indexes are physical and functional status classification tools, which, as noted, were not originally designed to predict outcome after major surgery.
Gilbert et al.44 performed a prospective cohort comparative evaluation of the ASA, Goldman, Detsky, and Canadian Cardiovascular Society (CCS) indexes for predicting the overall rate for cardiac complications in patients undergoing noncardiac surgery. Cardiac complications occurred in 6.4% of all patients. The areas under their respective receiver operating characteristic (ROC) curves for the different indexes were 0.625 (ASA), 0.642 (Goldman index), 0.601 (Detsky), and 0.654 (CCS). The authors concluded that existing indexes for predicting cardiac complications after noncardiac surgery perform better than chance, but no index is significantly superior to any other.
The ROC curve is a graphic representation of the relation between sensitivity and specificity in a given model. An important advantage of ROC analysis over traditional sensitivity and specificity analysis is that the area under the ROC curve is independent both of the cutpoint criteria chosen and the prevalence of the outcome of interest.45 This independence allows comparisons of ROC curves across study populations in which sensitivity and specificity would be distorted by differences in the prevalence of the outcome of interest. A model is considered perfect when the ROC area is 1.0, useless when it is <0.5 (i.e., under a line of no discrimination), has a low accuracy if it is between 0.5 and 0.7, and becomes useful with an area >0.7.
Dupuis et al.43 have developed a simple Cardiac Anesthesia Risk Evaluation (CARE) score based on clinical judgment and three clinical variables: comorbid conditions categorized as controlled or uncontrolled; surgical complexity; and urgency of the procedure. In a prospective observational study at a major university hospital, this clinical scoring system (CARE) was compared with the Parsonnet et al., Tuman et al., and Tu et al., multifactorial risk indexes,40,41,42 the three existing indexes for prediction of morbidity and mortality after cardiac surgery. The first 2,000 patients enrolled in this study served as a reference group to determine discrimination of each classification (CARE, Parsonnet index, Tuman index, and Tu index) using ROC curves (see Fig. 2.3). The next 1,548 patients were used to calibrate the scoring system. Areas under the ROC curves were similar for all the indexes. The authors concluded that the CARE score performs as well as the others for outcome prediction after cardiac surgery. They claim that the CARE score may be more appealing to health care providers because of its simplicity and focus on evaluation of the patient’s condition at a much more clinical level. Both Gilbert et al.44 and Dupuis et al.43 have shown that scoring systems based on subjective assessment may have a much wider acceptability in clinical use, because they are easier to use in everyday practice and are as accurate
in their ability to predict postoperative major complications as the more complex multifactorial risk indexes.
in their ability to predict postoperative major complications as the more complex multifactorial risk indexes.
 FIGURE 2.3 Receiver operating characteristic curves obtained with each risk model for prediction of mortality in the reference group (n = 2,000 patients). CARE, Cardiac Anesthesia Risk Evaluation; AUC, area under the curve. (From Dupuis JY, Wang F, Nathan H, et al. The cardiac anesthesia risk evaluation score: A clinically useful predictor of mortality and morbidity after cardiac surgery. Anesthesiology. 2001;94:194.) |
▪ RISK INDEXES VERSUS CLINICAL JUDGMENT
Are these risk indexes any more valuable than simple clinical judgment for both the patient and the anesthesiologist? Currently, the answer is not very clear. Most of the indexes were developed in a selected subset of patients. Therefore, they may lack general application and could lead to inconsistent results when applied to the population at large. The underlying assumption of all these indexes is that the specific variables studied are static. In other words, they cannot be modified with respect to their influence on the outcome. However in everyday practice, anesthesiologists, as perioperative specialist physicians, attempt to positively modify risk factors. Emphasis has now shifted from risk stratification to risk modification through interventions. If the interventions are successful, the risk factor decreases in importance and may not have the same negative impact on the outcome as compared to the situation when the severity of the risk factor is ignored.
The greatest value in developing these indexes may lie in the identification of important risk factors, which can then be targeted for intervention (i.e., through clinical judgment) to influence the outcome positively. Furthermore, perioperative management (postoperative care of the patient in an ICU rather than a standard room) can be tailored to the individual patient’s needs and risk stratification (again through clinical judgment). The other benefit of studying risk indexes is to make comparisons between groups (risk-adjusted measures). The rationale for risk adjustment is to remove one source of this variation, leaving residual differences in outcome to reflect the quality of clinical care given. Risk-adjusted outcome measures are also used to determine the appropriateness of care in a given situation or to produce “report cards” to compare the quality of service provided by institutions or individual health care providers.
How is Surgical Risk Evaluated?
Preoperative assessment of the patient traditionally concentrated on identifying and controlling any comorbid medical conditions and stratifying patients as low-, moderate-, or high-risk based on this approach.38,39 More recently, the practice has often evolved into performance of a series of invasive and noninvasive tests on patients at moderate or high risk for perioperative complications. For instance, in a patient with a history of cardiac disease, one aims to identify the location/region of ischemia and to quantify the myocardium at risk. In the process, the patient is triaged for surgery or for additional medical treatment and stabilization.46,47
▪ AMERICAN COLLEGE OF CARDIOLOGY/AMERICAN HEART ASSOCIATION GUIDELINES
The American College of Cardiology/American Heart Association (ACC/AHA) guidelines on perioperative cardiovascular evaluation for noncardiac surgery stress that preoperative testing should be limited to circumstances in which the results will affect patient treatment and outcome.47 They also state that the estimation of perioperative risk should integrate clinical determinants of risk (patient comorbidities), functional capacity (metabolic equivalent, MET or exercise duration), surgery-specific risk, and the results of stress testing (exercise electrocardiogram [ECG] testing, dipyridamole thallium imaging, or dobutamine stress echocardiography [DSE]).
Surgical procedures are often categorized as high-, intermediate-, or low-risk procedures on the basis of the hemodynamic stress they impose on the patient. High-risk surgery includes major abdominal and thoracic procedures, particularly in the elderly, and specifically, major vascular surgery, and prolonged surgery associated with major fluid shifts and/or blood loss. Intermediaterisk surgery includes orthopedic and prostatic surgery; low-risk surgery includes peripheral and laparoscopic procedures.
Ischemia versus Heart Failure
The current guidelines widely adapted in many institutions for perioperative workup of a patient with coronary artery disease undergoing noncardiac surgery (following the publication of ACC/AHA guidelines) are summarized in Table 2.2.
They are geared primarily towards workup for inducible ischemia rather than left ventricular dysfunction (heart failure). The ability of current noninvasive cardiac tests to evaluate heart failure is complicated by the overlap between coronary artery disease and heart failure. Furthermore, heart failure can result from a number of nonischemic causes (diastolic dysfunction, restrictive cardiomyopathy, asymmetric septal hypertrophy, etc.), in which case an ischemic workup will not provide any useful information. Preoperative heart failure has been identified as a major risk factor for other cardiac complications after surgery and is considered to be an important risk determinant in all the preoperative cardiac risk indexes.37,38,39,43 Perioperative heart failure is the most frequently encountered cardiac complication after noncardiac surgery.39,49
The reported incidence of postoperative heart failure in patients after major noncardiac surgery is 1% to 6%. However, this incidence increases to between 6% and 25% in patients with existing cardiac conditions such as coronary artery disease, prior heart failure, or valvular heart disease.50,51,52 The major determinant of perioperative morbidity and mortality is the inability of the heart to increase its output in response to surgical stress; this entity has been termed perioperative cardiac failure.50,51,52
Perioperative cardiac failure may only be clinically apparent in the postoperative period when oxygen demand is increased. It may also occur independently of both congestive cardiac failure and MI, although all three conditions may coexist. Therefore, preoperative evaluation should be focused on the detection of cardiac failure under stress. Cardiac failure in the elderly is frequently occult, because they often adjust their level of activity to their physical capacity. These patients may therefore be totally asymptomatic at rest for angina and heart failure. However, in the postoperative period when oxygen demand exceeds supply, this adjustment may not be an option. Hence, these patients are prone to cardiovascular complications.
Perioperative cardiac failure may only be clinically apparent in the postoperative period when oxygen demand is increased. It may also occur independently of both congestive cardiac failure and MI, although all three conditions may coexist. Therefore, preoperative evaluation should be focused on the detection of cardiac failure under stress. Cardiac failure in the elderly is frequently occult, because they often adjust their level of activity to their physical capacity. These patients may therefore be totally asymptomatic at rest for angina and heart failure. However, in the postoperative period when oxygen demand exceeds supply, this adjustment may not be an option. Hence, these patients are prone to cardiovascular complications.
TABLE 2.2 Perioperative Evaluation of Patient with Coronary Artery Disease for Noncardiac Surgery | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||
Echocardiography
Echocardiographic determination of resting left ventricular function may help predict adverse cardiac outcomes perioperatively. However, it does not appear to add important information to the clinical evaluation.53 In patients undergoing vascular surgery, there seems to be no relation between resting preoperative ejection fraction and postoperative MI or death.54 Given the limited evidence of demonstrable benefit of testing the resting left ventricular function by echocardiography, it is not currently recommended for routine preoperative left ventricular function evaluation.
What Are the Requirements for Preoperative Assessment?
A comprehensive preoperative evaluation should objectively assess the risks of a patient for cardiac and/or respiratory complications in the perioperative period. The testing methods used for the evaluation need to be noninvasive, reliable, reproducible, and valid at submaximal exercise capacity. To meet this end, there is a growing interest in looking at physiologic capacity (as measured by cardiopulmonary exercise testing [CPET] using exercise gas exchange data) as an objective method for predicting overall postoperative morbidity and mortality.52,55,56,57,58,59,60,61,62,63,64 The underlying principle is that the perioperative period imposes a degree of metabolic stress on the body similar to that of physical exercise, and similar demands are imposed on the heart, lungs, and the peripheral circulation system to support the body’s metabolic needs in both situations. This principle assumes that a patient’s capacity to increase oxygen delivery during exercise may correlate with his or her capacity to meet and sustain the increased metabolic demand on the body during the perioperative period.
▪ CARDIOPULMONARY EXERCISE TESTING
CPET is an objective method of evaluating cardiopulmonary function in which the patient exercises on a bicycle ergometer or, failing that, on a treadmill. During exercise, the inspired and expired gases from the patient are continuously sampled through a mouthpiece, and both oxygen uptake and carbon dioxide elimination are computed. Cardiac function is evaluated in terms of aerobic capacity (by gas exchange data), and the pulmonary function is evaluated by dynamic flow volume loops and [V with dot above]/[Q with dot above] measurements performed during exercise52 on a continuous basis throughout the duration of the test. The most important variables related to cardiopulmonary function (measured by CPET) that determine or predict postoperative morbidity and mortality are the anaerobic threshold (AT) and the peak oxygen consumption, [V with dot above]O2.
The advantage of using the AT over maximum aerobic capacity is that the AT does not rely on the motivation of the patient, and it occurs well before maximum aerobic capacity. Therefore, the AT is readily obtained even in elderly patients because it does not require great physical stress.
Older et al.65 previously have shown that cardiovascular mortality was restricted to patients with an AT of <11 mL/minute/kg. The combination of ECG evidence of myocardial ischemia and an AT indicative of moderate cardiac failure (i.e., <11 mL/minute/kg) was associated with a high mortality rate and the worst outcomes. Many elderly patients have a low AT without demonstrable ischemia. Aging results in reduced left ventricular compliance, making patients in this age group candidates for heart failure based on diastolic dysfunction. A low AT in the elderly, even in the absence of ischemia, is associated with a high mortality risk. Therefore, solely relying on symptoms of angina or the results of an ECG may not accurately predict cardiac risk in this set of patients.
Older et al.65 previously have shown that cardiovascular mortality was restricted to patients with an AT of <11 mL/minute/kg. The combination of ECG evidence of myocardial ischemia and an AT indicative of moderate cardiac failure (i.e., <11 mL/minute/kg) was associated with a high mortality rate and the worst outcomes. Many elderly patients have a low AT without demonstrable ischemia. Aging results in reduced left ventricular compliance, making patients in this age group candidates for heart failure based on diastolic dysfunction. A low AT in the elderly, even in the absence of ischemia, is associated with a high mortality risk. Therefore, solely relying on symptoms of angina or the results of an ECG may not accurately predict cardiac risk in this set of patients.
CPET is a totally noninvasive test that is quick, cheap, easy to perform, and requires no special preparation. It is promoted as being able to objectively evaluate the extent of any cardiac failure and/or myocardial ischemia, provide insight into stroke index and the presence of pulmonary artery hypertension, and define obstructive and restrictive lung disease and [V with dot above]/[Q with dot above] inequality better than conventional preoperative respiratory function tests.52 The challenge still remains to incorporate risk evaluation to positively change outcome through appropriate perioperative measures.
The American Thoracic Society/American College of Chest Physicians’ joint statement on the use of CPET is as follows, “Indications for cardiopulmonary exercise testing as a preoperative assessment tool, is encouraged for lung cancer surgery, lung volume reduction surgery and evaluations for both lung and heart transplantation.” In a section titled “Preoperative Evaluation for Other Procedures” it states, “Work has shown that cardiopulmonary exercise testing is helpful in objectively assessing the adequacy of cardiovascular reserve and in predicting cardiovascular risk in elderly patients.”66 It remains to be seen if this preoperative testing will be widely accepted and adopted in the future.
▪ BRAIN NATRIURETIC PEPTIDE
Strong evidence exists now for the use of brain natriuretic peptide (BNP) in the diagnosis of acute heart failure and for improving clinical outcomes with a BNP-guided approach to heart failure care. In the setting of acute breathlessness, plasma BNP enabled the discrimination of cardiac causes from noncardiac causes of dyspnea with high accuracy.67 In a smaller study of patients referred for evaluation for cardiac transplantation, plasma NT-pro-BNP was able to predict the combined endpoint of death and the need for transplantation better than several key prognostic indicators, including left ventricular ejection fraction (LVEF), peak oxygen consumption ([V with dot above]O2), and the heart failure survival score.68 In patients with heart failure having an LVEF of ≤45%, the combination of plasma BNP and percentage of maximal predicted [V with dot above]O2 attained on cardiopulmonary testing predicted cardiovascular mortality similarly.69 The combined measurement of plasma BNP and troponin may have greater prognostic utility than measurement of either marker alone.70
Postoperative plasma BNP and cardiac troponin I levels have been shown to be independent predictors of postoperative cardiac dysfunction after cardiac surgery.71 This prospective observational study in 92 consecutive patients in a university hospital also showed that simultaneous measurement of BNP and cardiac troponin I improved the risk assessment of postoperative cardiac dysfunction. However, the association between BNP levels and 1-year outcome was no longer significant after adjusting for LVEF. The clinical utility of these newer biomarkers in the management of heart failure and other causes of cardiac dysfunction merits further study through carefully designed prospective investigations. It is likely that, in the future, the use of multiple biomarkers in combination may improve the diagnosis, treatment, and risk stratification of patients with heart failure.72 However, to date there is no data on the use of biomarkers as reliable tools for predicting postoperative morbidity and mortality or as a measure for preoperative risk stratification.
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
