Since the 1960s, the desire to find the optimal tool of risk assessment has been present. The American Society of Anesthesiologists (ASA) developed the Dripps Index as a way not only to identify risk among patient groups, but also to provide a common framework and communication device that could easily be distributed among differing medical specialties [14]. In 1970, Vacanti et al. [15] used the index to predict cardiac death within 48 hours of surgery. Within the five physical status
grades identified, perioperative mortality rates range from 0% for ASA status 1 to 9.4% for ASA status 5. However, some of the major drawbacks to the utilization of the ASA score are that it was developed prior to multivariate clinical prediction rules, has limited utility, is very subjective, and is not uniformly reproducible [16].

One of the original cardiac risk assessment tools developed in the 1970s by Goldman was an elaborate attempt to identify those patients at undue risk [17]. Risk assessment was based on several clinical variables. Goldman identified nine independent variables associated with perioperative cardiac events. These are included in Table 149.2, and consist of variables ranging from advanced age to the presence of significant valvular heart disease.

Each variable was assigned specific points and the patients were divided into risk class depending on the number of points generated. The highest classification—class IV (more than 26 points) was associated with a 78% incidence of major cardiac complications in the perioperative period. However, the drawback to use of the tool was the cumbersome nature, making the utilization of the Goldman risk assessment tool somewhat impractical.

In 1986 Detsky attempted to modify the Goldman risk assessment tool by the addition of angina severity and a history of recent pulmonary edema [18]. Broad categories included the variables of coronary artery disease, Canadian Cardiovascular Society Angina Classification, alveolar pulmonary edema, suspected critical aortic stenosis, arrhythmias, poor general medical status, emergency surgery, and age 70 or older. However, just as with Goldman, this risk assessment tool was viewed to be exceedingly cumbersome. It appears that both indices may not have sufficient discriminate power to identify significant coronary artery disease in patients at the lower end of the spectrum of clinical risk [19] and both indices have been refuted or supported by an equal number of studies [20].

Adding to the controversy has been a prospective cohort study that compared the varying risk indices for patients undergoing noncardiac surgery. Gilbert et al. [16] compared 2,035 patients referred for consultation prior to noncardiac surgery and four risk indices: the Dripps Index of the ASA, the original cardiac risk index described by Goldman, the modified Detsky (which had been modified in 1997 by the American College of Physicians by stratifying patients into three risk groups) [21], and the Canadian Cardiovascular Society (CCS) Index for angina level [22]. The most striking finding of the study was that existing cardiac risk prediction methods had a generally poor degree of accuracy.

Eagle et al. [23], while assessing the validity of dipyridamole-thallium stress testing in vascular patients, developed another set of risk criteria for patients undergoing major vascular surgery. The group found five clinical predictors of postoperative cardiac events. These included: presence of Q waves on resting ECG, history of angina, history of ventricular ectopy requiring treatment, diabetes mellitus requiring medical treatment, and age above 70 years. Also on logistic regression, the group noted two independent dipyridamole thallium test predictors of ischemic events that included thallium redistribution and ischemic ECG changes during or after pharmacologic stressing.

In an attempt to simply the Goldman index, Lee et al. [24] developed the Revised Cardiac Risk Index (RCRI) Stratification System. The RCRI for the first time identified six independent risk predictors associated with cardiac morbidity and noncardiac surgery. These included: high-risk surgery (examples included intraperitoneal, intrathoracic, or suprainguinal vascular reconstruction), a history of ischemic heart disease (excluding previous revascularization), a history of congestive heart failure (CHF), a history of cerebrovascular disease, preoperative treatment with insulin, and a preoperative serum creatinine level more than 2.0 mg per dL (greater than 177 μmol per L).

Cardiac events were determined to be myocardial infarction, cardiac arrest, pulmonary edema, or complete heart block. Four classifications were noted in which risk factors ranged from 0 to 3 or more and correlated to event rate:

The RCRI has been the most widely accepted risk index, and Romero and de Virgilio [20] have proposed utilizing the RCRI to identify patients who should be treated with strategies to reduce oxygen consumption rather than undergo additional noninvasive testing. They based their recommendations on comments elicited by Bodenheimer [25], who felt that improved outcomes were more likely a result from controlling postoperative myocardial oxygen demand than additional risk stratification.

Other attempts at risk stratification and adjustment are mentioned in the literature. In 2004, Atherly et al. [26] compared the National Surgical Quality Improvement Program (NSQIP), the DxCG, and the Charlson Comorbidity Index. The NSQIP [27] is based on a medical record abstraction of 45 preoperative and 17 intraoperative factors. Factors are multiplied by weights drawn from a model developed using 41,360 patients from the Veteran Affairs Health Care System. Some of the major components of the NSQIP specific to mortality include: ASA class, ventilator dependence, emergency case, age, abnormal albumin, ascites, complexity score, and contaminated wound [28]. In addition to those mentioned earlier, functional status, a history of chronic obstructive pulmonary disease, anemia (hematocrit 38% or less), and elevated white blood cell counts (11,000 or more) are important predictors of morbidity. The ultimate risk score represents the probability of individual patient mortality.

The DxCG uses International Classification of Disease (ICD-9) codes, sex, and age to assign a continuous risk score, and the Charlson Comorbidity Index (CCI) was developed to predict empirically the probability of 1-year mortality. The CCI contains 19 categories of comorbidities drawn from the ICD-9 codes. Each of the categories has a weight, which indicates an increase in the risk for 1-year mortality and scores range from 0 to 6.

Atherly et al. [26] found substantial disagreement in the risk assessment calculated by the three methodologies. A weak association was noted between the CCI and DxCG, but neither correlated well with the NSQIP. Overall, the NSQIP was felt to be the best predictor of surgical mortality.

One of the major changes in the 2007 revision of the ACC/AHA guidelines is the manner in which risk is assessed. In the 2002 version of the guidelines, risk factors were divided into three groups: major, intermediate, and minor clinical predictors [29]. With the new revision, the minor clinical predictors were removed from the algorithm because, although they may signify risk for coronary disease, they have not been shown to independently increase risk for perioperative cardiac complication [30].

Also changed in 2007, the major clinical predictors have been renamed active cardiac conditions (Table 149.3). Because of the increasing use of the Revised Cardiac Risk Index created by Lee et al. [24], the committee chose to replace the intermediate clinical predictors with five of six risk factors identified by Lee’s group. These five risk factors are: history of ischemic heart disease, compensated heart failure, history of cerebrovascular disease, diabetes mellitus, and renal insufficiency. The sixth risk factor identified by Lee et al., type of surgery, is addressed elsewhere in the new guidelines.

The guidelines also focused significantly on the concept of functional capacity. Functional capacity is best expressed in metabolic equivalent (MET) levels that correlate with specific activities. Basic energy expenditure for activities of daily living (e.g., eating, walking) are around 1 to 4 METs, while strenuous exercise is often more than 10 METs [31]. It has been shown in prior studies that patients unable to obtain a 4-MET demand do poorly in the perioperative period [32] as well as in the long term [33].

Different surgical procedures are clearly associated with varying amounts of hemodynamic stress. For example, application and release of an aortic cross clamp during abdominal aortic aneurysm repair induces far more physiologic insult than
cataract surgery does. Furthermore, recent evidence suggests that major vascular surgery (excluding carotid endarterectomy) may be associated with more than 5% risk for perioperative cardiac death or nonfatal myocardial infarction [30]. With this in mind, the most recent revision of the ACC/AHA guidelines classifies vascular surgery separately as the highest risk group [30] (Table 149.4). Procedures associated with a 1% to 5% cardiac risk, such as orthopedic and intraperitoneal surgeries, are classified as intermediate risk. Most ambulatory surgeries are associated with less than 1% cardiac risk and are classified as low risk.

In the 2007 revision, the authors generated a five-step algorithm for preoperative risk assessment (Fig. 149.2). This was a definite improvement from the somewhat confusing 3-part, 8-step algorithm published in 2002. The simplified recommendations were necessary considering the abysmal (as low as 21%) implementation of the 2002 guidelines [34]. These new guidelines reflect the authors’ sentiment in their opening statement that cardiac intervention is not indicated unless it would be performed regardless of a preoperative context. In addition, the algorithm offers recommendations for noninvasive testing and treatment with beta-blockers for selected patients.

Despite these improvements, many authors are still critical of the algorithm. Brett argues that the guidelines are still too ambiguous, referring to the final point of the decision tree: “consider testing if it will change management” [35]. He also makes a point that sometimes noninvasive testing helps patients weigh the risks and benefits of truly elective surgery. In any case, the new algorithm will likely decrease the number of noninvasive test ordered, thus reducing cost and delay in performing elective procedures.

Not long ago it was commonplace to see electrocardiograms in the chart for most surgical patients as part of a preoperative workup. Because these extensive workups were often fruitless, and some testing caused more harm than good, the ASA assembled a task force to develop a practice advisory for
preanesthetic evaluation [36]. The task force cited that few screening ECG findings resulted in changes in clinical management. They also stated that based on evidence, age alone may not be an indication for ECG. Proponents of screening ECGs argue that these studies may identify patients with coronary disease not recognized by clinical history. Moreover, these newly identified patients could then be further tested or medically managed with beta-blockade. However, this argument may be flawed for several reasons. First of all, a positive ECG in an asymptomatic patient would not alter further testing if the practitioner uses the ACC/AHA algorithm [30]. Second, according to van Klei et al., ECG abnormalities, including left and right bundle branch blocks, were no more predictive of postoperative MI than history alone [37]. Finally, starting beta-blocker therapy is probably not indicated in otherwise asymptomatic patients [30]. Fleisher, however, does make one argument that may be valid for obtaining preoperative ECG [38]. Without a preoperative ECG, the first occasion that the ECG may be seen as abnormal is when the patient is in the operating room prior to induction. Under these circumstances, it may be beneficial to compare the new findings with an old ECG to identify the acuity of the changes and determine whether or not to proceed. Currently, however, the ACC/AHA states that preoperative screening ECGs are indicated only for vascular surgeries and for certain patient populations having intermediate-risk surgery (Table 149.5) [30].

Exercise stress testing is a well-established mechanism of assessment that allows the identification or absence of myocardial ischemia while the patient is undergoing physical exertion. The purpose of the examination is to elevate the myocardial oxygen consumption to a rate in which demand outweighs supply, leading to ischemic changes on ECG. The inherent drawback of this method of assessment is that it relies on patient participation. At times, due to deconditioning or medical issues, such as claudication, the patient cannot reach target heart rate and thus ischemic episodes may be missed.

Unfortunately in meta-analysis, the mean sensitivity of exercise ECG testing for the prediction of multivessel coronary artery disease has been reported to be 81% (range 40% to 100%) with a mean specificity of 66% (range 17% to 100%) [41]. The meta-analysis also reconfirmed that the sensitivity of the examination was adversely affected in patients who could not reach maximal heart rate, especially vascular surgery patients in which approximately 50% could not reach the target rate.

In addition to the failure to reach target heart rate, other limitations of exercise testing exist. These include ECG changes on resting ECG, the presence of left bundle branch block, failure in determining the extent of myocardial ischemia, and lack of information regarding left ventricular function [42].

To overcome some of the inherent problems of exercise stress testing, pharmacologic stress myocardial perfusion imaging was developed [43]. This examination consists of the administration of a vasodilating agent such as adenosine or dipyridamole to induce vasodilation that would parallel the effect of exercise on coronary anatomy. In addition, a radionuclide is administered, such as thallium-201. Images are obtained over time and positive examinations are those in which areas of initial filling defects resolve, or undergo redistribution of thallium, during the rest phase.

Several complications and contraindications exist with the use of adenosine and dipyridamole. Since they are potent vasodilators, these agents are obviously contraindicated in those patients with preexisting hypotension and ongoing symptoms of unstable angina. Other relative contraindications to administration of adenosine include high-degree atrioventricular block, bronchospastic disease, and atrial arrhythmia disorders such as sick sinus syndrome.

Eagle et al. found that patients with one or two risk factors for coronary artery disease, and redistribution on dipyridamole thallium had a 29% cardiac event rate versus a 3.2% rate in patients without redistribution. The sensitivity of the examination, however, appears to be in detecting the presence or absence or coronary artery disease, not ischemia [23].

In addition it has been reported that the accuracy and positive likelihood ratio for dipyridamole thallium stress testing is low and that the examination does not provide independent prognostic value beyond clinical risk stratification [44]. Other prospective blinded studies confirmed a lack of association between reversible defects on dipyridamole thallium and adverse cardiac events in patients undergoing elective vascular surgery (of note, these studies excluded low-risk patients undergoing vascular surgery) [45,46].

In the study by de Virgilio et al. [46], the adverse cardiac event rate was 13.8% for patients with a reversible defect on thallium testing versus 9.8% for those who did not have a reversible defect (p = 0.70). The adverse event rate in patients with two or more reversible defects was 12.5% versus 11.1% in patients with fewer than two reversible defects. Sensitivity with two or more defects was 11%, with a specificity of 90%. The overall positive and negative predictive values were 12.5% and 89%, respectively. The authors concluded that since there was no demonstrable correlation between dipyridamole thallium and perioperative adverse cardiac events, one could not recommend the test as a screening tool prior to vascular surgery.

Another imaging study is dipyridamole technetium-99m sestamibi testing. Technetium-99m sestamibi is a radiotracer that differs from thallium-201 and ultimately allows for acquisition of higher resolution tomographic cardiac images. Stratmann et al. [47] studied 229 patients scheduled for vascular surgery who underwent sestamibi testing. Of those enrolled, 197 underwent surgery within 3 months of the initial examination with an overall cardiac event rate of 5%. The perioperative cardiac event rate between those with normal, abnormal, or reversible sestamibi images was not clinically significant; however, abnormal and reversible sestamibi images were independent multivariable predictors of increased risk of late cardiac events.

Dobutamine stress echocardiography (DSE) was developed as a tool for assessing the presence of coronary artery disease and
was reported by Berthe et al. [48] in 1986. Essentially the examination is composed of the administration of a pharmacologic inotropic agent (e.g., dobutamine), which is designed to increase heart rate and myocardial contractility, thus increasing myocardial oxygen consumption. In the presence of coronary artery disease, demand will overcome supply and myocardial dysfunction will be present. Myocardial dysfunction will be evident by echocardiography, manifested by areas of hypokinesis, akinesis, or dyskinesis. The development of new wall motion abnormalities following dobutamine administration is considered an indication of significant coronary artery disease [49].

When dobutamine stress echocardiography and dipyridamole-thallium testing were compared in the same patient population, they appeared to have comparable specificity and sensitivity [50]. A subsequence meta-analysis study revealed a 9% incidence of perioperative myocardial infarction in patients with reversible ischemia or regional wall abnormalities in one or more areas [51].

The Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography (DECREASE) Study Group performed a large retrospective study with results released in 2001. The study noted that the adverse event rate was 10.6% in patients with three or more cardiac risk factors and five or more segments of new wall motion abnormalities (NWMAs) versus a 2% adverse event rate in patients without NWMAs. It is also interesting to note that the study reported perioperative death and myocardial infarction rates of 6.5%, 10%, and 16% in patients with respective scores on a modified Revised Cardiac Index of 3, 4, and 5 who were treated with beta-blockade but also had ischemia on DSE [52]. A drawback to the utilization of echocardiography was that the study showed that DSE did not add incremental value in low- or medium-risk patients (score of 0 to 2 on Revised Cardiac Risk Index) [3].

Although the results of this retrospective study were encouraging, there are other studies that tend to question the validity of DSE for preoperative evaluation. It appears that echocardiography has limited prognostic value as a routine test. Rohde et al. [53] reported that an abnormal echocardiogram with any degree of systolic dysfunction, moderate to severe left ventricle hypertrophy, moderate to severe mitral regurgitation, or aortic gradient of 20 mm Hg or higher provided a sensitivity of 80%, specificity of 52%, positive predictive value of 12%, and negative predictive value of 97%. However, severe left ventricular (LV) dysfunction compared to mild–moderate LV dysfunction did not have a strong association with cardiogenic pulmonary edema and MI. Thus, given the heterogeneity of findings, it appears that echocardiography adds little to risk models.