Uncertainty about the true incidence of perioperative infarction has important consequences. Many patients facing noncardiac surgery undergo extensive testing to assess their risk for having a cardiac complication. In patients perceived to be at high risk, invasive procedures ranging from pulmonary artery catheterization [13] to prophylactic coronary artery bypass [14-16] are recommended to reduce the risk. However, if the true incidence—and hence a given individual's risk—for perioperative infarction is not known, we cannot be sure that the interventions we recommend do not carry risks that outweigh the adverse consequences of a perioperative infarction with the planned procedure.

When we began this study, the most commonly used approach to estimating an individual's risk for perioperative cardiac complications was that of Goldman and colleagues [5] (later modified by Detsky and coworkers [10]). Goldman's cardiac risk index is based on nine differentially weighted factors and predicts the likelihood that any of three complications—infarction, pulmonary edema, and ventricular tachycardia—may occur. Grouping these together limits the index's usefulness. They differ somewhat etiologically, require different approaches for detection, are treated differently, and have different implications for longevity and functional status. More importantly, they require different approaches to prevention.

We thought that the debate about prevention of perioperative infarction needed more precise data on incidence, and our main objective in performing the present study was to provide that data. We also wanted to evaluate the performance of a preoperative risk stratification scheme that classified patients according to their probability of having coronary artery disease and to determine whether other clinical factors in addition to coronary artery disease are associated with an increased risk for perioperative infarction.

Methods

Top Methods Results Discussion Author & Article Info References

This study was performed at the Houston Veterans Affairs Medical Center from July 1987 through December 1989. The participants were 1487 consecutive men age 40 years or older who had major elective or urgent noncardiac surgery. Women were excluded; they account for less than 1% of operations at our hospital. The study was approved by the relevant institutional review boards.

Three kinds of patients were studied: active participants, who consented to a preoperative and postoperative evaluation protocol (n = 512); chart review patients, who either declined the evaluation protocol (n = 184) or had surgery before they could be approached about participating (n = 139); and negligible-risk patients (n = 652) whose risk for perioperative infarction was extremely low. Because of resource constraints, negligible-risk patients were not asked to participate, and the only postoperative data collected on them were the occurrence and cause of death.

Preoperative Risk Stratification

The risk stratification scheme tested prospectively in this study is based on the premise that because autoregulatory mechanisms are able to maintain myocardial perfusion in the normal heart over a wide range of conditions, coronary atherosclerosis is a necessary precondition for perioperative infarction. Information from the history, examination, and clinical record allows the scheme to assign patients preoperatively into four risk strata for perioperative infarction, representing different levels of prevalence of coronary atherosclerosis, taking into account that significant coronary atherosclerosis can exist in a subclinical form. We projected that the incidence of infarction would decrease down the strata as the prevalence of coronary atherosclerosis decreased because a smaller proportion of the stratum would be at risk.

The high-risk stratum included patients almost certain to have coronary atherosclerosis: those with a history of infarction or electrocardiographic (ECG) evidence of such according to the Minnesota Code [17], typical angina (in whom the likelihood of coronary disease is 90% [18]), angiographically documented significant coronary disease (> 50% reduction in the left main or a 70% or more reduction in the right, left anterior descending, or left circumflex arteries), or a history of coronary bypass surgery. The intermediate-risk stratum included patients without evidence of coronary atherosclerosis but with atherosclerosis elsewhere (the prevalence of significant occult coronary disease in such patients is about 30% [14]): those with a history of stroke or transient ischemic attacks, a history of peripheral vascular surgery for atherosclerosis (or whose scheduled operation was such), claudication, or carotid bruit. Patients with atypical chest pain were also placed here because about half of these patients have coronary disease [19]. The low-risk stratum included patients who had no clinical atherosclerosis but who had a high atherogenic risk factor profile (sex, age, cigarette smoking, blood pressure, cholesterol, glucose tolerance, and ECG abnormalities), that is, at least a 15% likelihood of a cardiac event within 6 years, based on the Framingham probability tables [19] in the Coronary Risk Handbook [20]. A small proportion of such patients had significant, although occult, coronary disease [21]. We also placed in the low-risk stratum men older than 75 years without clinical atherosclerosis regardless of their likelihood of a future cardiac event. Patients who met none of the above conditions were, by definition, in the negligible-risk stratum. Using a process of elimination from high- to negligible-risk, stratifying a patient took about 10 minutes. Only one criterion was necessary to place a patient in a stratum. Figure 1 illustrates how the patients were distributed among the strata.

View larger version (29K): [in this window] [in a new window]   Figure 1. Study population. Chart review patients are patients who declined to undergo a structured preoperative evaluation and postoperative testing for infarction or who were missed by the study team.

Preoperative Evaluation and Intraoperative and Postoperative Data

Active participants had a preoperative history, cardiovascular system review, and examination by an investigator. A standard questionnaire was used to detect angina and intermittent claudication [22], and standardized questions were used to detect transient ischemic attacks. Serum total creatine kinase and its MB isoenzyme were measured, and an ECG was obtained. No other preoperative cardiac tests were part of the study protocol. Standard cardiac tests were obtained at the discretion of the surgeon, but during the study years dipyridamole thallium scintigraphy was not performed at our hospital. Chart review patients were slightly older and had a higher preoperative Goldman cardiac risk index [5], but they had a lower prevalence of coronary disease (Table 1).

Active participants had an ECG, total creatine kinase, and MB measurement between 0600 and 0800 on the first three postoperative days. Operative data were obtained from anesthesia records. Sustained hypotension (a 30% decrease in the mean arterial pressure sustained for more than 10 minutes) was noted because it is associated with perioperative infarction [4, 6, 8, 23]. Data on postoperative cardiovascular problems were obtained from the chart and, in active participants, from investigators' visits.

Total creatine kinase values were quantitated enzymatically and the MB isoenzyme was quantitated immunochemically using Roche reagents (Roche Diagnostic Systems, Nutley, New Jersey). Creatine kinase levels are elevated by surgical skeletal muscle trauma, but the MB isoenzyme retains specificity for myocardial damage after noncardiac surgery [24]. In patients with myocardial necrosis, creatine kinase MB almost always remains elevated for more than 24 hours [25]. A certified coder analyzed all ECGs using the Minnesota Code [18], a standardized system of waveform measurement and classification that controls interobserver variability and interpretation bias. A bias exists toward underdiagnosis of infarction in chart review patients because we had to rely on diagnostic tests ordered at the surgeon's discretion.

Diagnostic Criteria for Perioperative Myocardial Infarction

"Perioperative" denotes the time from arrival in the operating suite to hospital discharge. A diagnosis of infarction was based on new postoperative Q-QS patterns (62% sensitivity and 88% specificity for infarction [26]) and serial changes in creatine kinase MB levels (the immunochemical assay has 94% sensitivity and 95% specificity for acute infarction [27]). Patients whose ECGs or enzymes were equivocal underwent technetium pyrophosphate infarct scintigraphy. Infarction was diagnosed in four patients with equivocal ECGs or enzymes whose scintigram was positive using the criteria of Righetti and colleagues [28].

Statistical Analysis

The association between potential predictors and perioperative infarction was explored using the chi-square test and relative risk for categorical variables and the t-test for continuous variables. Logistic regression was used to test for independence among predictors. Highly correlated variables were not included in the same model. Because our aim was to create a clinical prediction model that would be helpful to clinicians performing preoperative assessments, only variables that could be known preoperatively were tested in the models. Variables were selected for testing based on theoretical connections with infarction rather than exclusively on univariate significance levels. Because of the low event rate, a level of 0.15 was used to remove variables from the models. Goodness of fit was assessed using the log likelihood chi-square statistic [29].

Because the results of the analysis of active participants alone are similar to those of the combined group of active participants and chart review patients, the groups are combined in the analysis.

Results

Top Methods Results Discussion Author & Article Info References

Incidence of Myocardial Infarction

The overall incidence of perioperative myocardial infarction in active participants was 1.7% (8 of 512; 95% CI, 0.4% to 2.7%). Despite the potential bias toward underdiagnosis of infarction in chart review patients, incidence in this group was 2.2% (7 of 323; CI, 0.4% to 3.9%). Incidence in both groups combined was 1.8% (15 of 835; CI, 0.8% to 2.8%). The incidence of myocardial infarction and other perioperative complications is given in Table 2.

The preoperative risk stratification scheme partitioned the patients into groups with a decreasing incidence (Table 3). Patients with coronary artery disease (the high-risk stratum) had a 4.1% incidence (13 of 319; CI, 1.8% to 6.4%). Patients with peripheral vascular disease but no evidence of coronary atherosclerosis (intermediate-risk) had a 0.8% incidence (2 of 260; upper bound of CI, 2.0%). No infarctions were observed in 256 patients who had high atherogenic risk factor profiles but no clinical atherosclerosis (the low-risk stratum, upper bound of CI, 1.2%). No cardiac deaths occurred in the 652 negligible-risk patients who had no clinical atherosclerosis and low atherogenic risk factor profiles.

None of the 15 patients with infarctions had unstable angina preoperatively or had surgery within 3 months of a previous infarction. Ten of the 15 infarctions were diagnosed on or before the third postoperative day; three were diagnosed on day 4, one on day 7, and one on day 19. Chest pain was a presenting symptom in only 2 of 15 patients (13%). Three of the 15 (20%) were completely asymptomatic; they would have been missed without the protocol ECGs and enzyme tests. Four (27%) myocardial infarctions were fatal.

Risk Factors for Infarction

The stratification scheme assessed the importance of coronary disease as a risk factor for perioperative infarction. We evaluated other potential risk factors in the combined group of low-, intermediate-, and high-risk patients. Univariate comparisons Table 4 indicate that besides coronary disease, a history of heart failure, previous vascular surgery, and signs of heart failure on the preoperative examination were significantly associated with infarction. Men who had intraoperative hypotension or who had vascular surgery were more likely to have perioperative infarction. Operation time was shorter in men who had infarction, which was due to the disproportionate number of amputations (short procedures) in this group.

We used multivariable analysis to test age, race, coronary artery disease, history of myocardial infarction, diabetes, hypertension, stroke, angina, heart failure signs on the preoperative examination, chronic use of ß-blockers, a planned vascular procedure, and planned general anesthesia. Four were found to be independently associated with infarction Table 5: age more than 75 years, the presence of heart failure signs preoperatively, coronary artery disease, and a planned vascular procedure.

No infarctions occurred in the 256 low-risk stratum patients. Summary data indicate a fairly high prevalence of chronic medical problems in the stratum, although they seemed not to increase the risk for perioperative infarction. The mean age was 65.5 years ± 6.97 (SD); 52% were white; and 64% were smokers. Five percent had a history of heart failure (from nonischemic causes), and 6% had evidence of heart failure on the preoperative examination. Thirty-seven percent had hypertension, 16% had diabetes, and 36% had another chronic medical illness. The group's average likelihood of a cardiac event within 6 years according to the Framingham risk Equation was 20% ± 4%.

Discussion

Top Methods Results Discussion Author & Article Info References

We found the overall incidence of perioperative infarction in men older than 40 years undergoing nonemergent noncardiac surgery to be very low (1.8%). Incidence varied, however, according to the prevalence of coronary disease in subgroups of study patients. The incidence of perioperative infarction in men with obvious coronary atherosclerosis was 4.1%; it was 0.8% in men without signs of coronary disease but evidence of atherosclerosis in other circulatory beds. These figures are comparable to those of a recent study at another Veterans Affairs Medical Center [7]. Methodologic differences make comparisons with earlier studies difficult, but it appears that infarction in patient groups with a high prevalence of coronary artery disease has declined over the past decade: Incidence in most earlier studies exceeds 4% [2, 4, 8, 23, 30]. On the other hand, our study corroborates the low (under 1%) incidence found in most studies of groups with a low prevalence of coronary atherosclerosis [1, 2, 4].

The overall performance of our risk stratification scheme supports the proposition that coronary disease is the most powerful risk factor for perioperative infarction. This rapid, inexpensive scheme distinguished different levels of risk for perioperative infarction and may be useful in deciding which patients need further evaluation. Coronary disease prevalence, and therefore the risk for perioperative infarction, is negligible in patients who have neither clinical atherosclerosis nor high atherogenic risk factor profiles. Workup of such patients beyond the history and physical examination would not be cost-effective. Neither does a high atherogenic risk factor profile warrant extensive preoperative cardiac evaluation: In 256 such patients in our study, no infarctions occurred. Almost all the infarctions in this study occurred in men whose coronary disease was obvious during the preoperative workup. The rest occurred in men who had peripheral vascular disease without evident coronary disease. About one third of the latter can be expected to have clinically significant occult coronary disease [14]. We found the incidence of perioperative infarction in such patients to be 0.8%. Whether this incidence is high enough to warrant preoperative cardiac workup in these patients requires a risk–benefit analysis beyond the scope of this article. Other data from this study can be used to help decide which patients with coronary or peripheral atherosclerosis would benefit most from additional preoperative testing. Age more than 75, preoperative evidence of heart failure, and a planned vascular operation were independently associated with an increased risk for infarction. Heart failure is a predictor of perioperative complications in some [5] but not all studies [7]. Control of symptoms and signs of heart failure preoperatively is a desirable goal because there are physiologic reasons to believe that doing so may reduce risk for perioperative myocardial infarction. Our finding that patients undergoing vascular surgery are at increased risk for peri-operative infarction is consonant with a growing body of literature [9, 12, 15, 31, 32]. In this study, men with overt coronary disease had 2.7 times the risk for infarction if their operation was a vascular procedure (CI, 0.86 to 8.70; data not shown). Overall, the adjusted odds ratio for vascular surgery compared with nonvascular surgery was 4.3.

This study has several limitations. The preoperative assessment was as standard as possible, but the diagnosis of coronary disease is always imprecise when made clinically. However, our aim was to replicate the circumstances under which preoperative risk evaluation is done in actual practice. Our decision to include chart review patients in the analysis may be challenged on the grounds that substantial misclassification may have occurred on both the risk factors and outcome. On the other hand, we thought that, because of the healthy volunteer effect, excluding them might give an overly optimistic picture of the incidence of infarction. In fact, the incidence was similar in chart review patients, despite the bias toward underdiagnosis. The results of the study should be generalized with caution because we studied only male veterans undergoing surgery at a large urban Veterans Affairs hospital. Such patients may have, on average, poorer health and a higher prevalence of harmful habits. In addition, the clinical prediction logistic model has not been validated prospectively.

Many questions remain about perioperative infarction but the most pressing pertain to risk assessment and prevention. On the basis of our findings, preoperative testing beyond the history, physical examination, and ECG would not have been cost-effective for men in the negligible-, low-, and, some would say, the intermediate-risk strata. Such patients account for 61% (79% if the intermediate stratum is included) of men older than 40 years undergoing major noncardiac surgery at our hospital. But once a patient is classified as high risk, what is the safest, most effective way to reduce the likelihood of perioperative infarction? None of the proposed risk-reducing maneuvers, from the intraoperative use of certain drugs, intraoperative pulmonary artery catheterization, postoperative intensive care, to coronary artery bypass, has been evaluated in a randomized trial. If the incidence of infarction is truly as low as we found in this study, some of the alleged risk-lowering interventions in use today may in fact be more dangerous than doing nothing.