The Preoperative Electrocardiogram: Have We Reached the End of an Era?

William Einthoven first recorded the electrical activity of the heart 100 years ago,1 and before long the electrocardiogram (ECG) became a standard tool for diagnosing a wide range of cardiac conditions, including acute and chronic myocardial ischemia and infarction, left and right ventricular hypertrophy, various congenital heart diseases, and disturbances of cardiac rhythm and conduction, and an array of noncardiac conditions, such as electrolyte disorders (especially hyperkalemia and hypocalcemia), drug toxicities (e.g., digoxin, antiarrhythmic agents, antidepressants), pulmonary embolism, and even hypothermia. Moreover, the importance of the ECG in clinical practice has not declined despite the development of vastly more sophisticated, more accurate, and, yes, more expensive techniques for diagnosing cardiac disease (echocardiography, cardiac catheterization, magnetic resonance imaging, and positron emission tomography). Given the wealth of information provided by the standard 12-lead ECG at a remarkably low cost (the current Medicare reimbursement rate for an ECG is approximately $24, including the professional fee), it is not surprising that the ECG has become an established component of the routine clinical assessment for cardiovascular disease in inpatient and outpatient settings. In particular, because perioperative cardiac complications are common and have been associated with substantial morbidity and mortality,2,3 acquisition of a preoperative 12-lead ECG is standard practice in the United States, especially for older adults undergoing major surgical procedures. But now, in this era of cost-containment and evidence-based medicine, many time-honored medical traditions have come under closer scrutiny, and, in this issue of the Journal of the American Geriatrics Society, Liu et al. challenge the utility of the preoperative ECG in geriatric patients undergoing a broad range of major noncardiac surgical procedures.4 In 1977, Goldman et al. published the results of a landmark study in which 1,001 patients aged 40 and older undergoing major noncardiac surgical procedures were evaluated prospectively.5 In this study, the authors identified independent risk factors for serious perioperative cardiac complications, defined as new myocardial infarction (MI), heart failure, ventricular tachycardia, or cardiac death. Nine independent risk factors were identified. Among these were the presence of nonsinus rhythm or ectopic atrial beats on the preoperative ECG, frequent ventricular premature beats, and age 70 and older; these three risk factors constituted the third-, fourth-, and fifth-strongest independent risk factors for major cardiac events. In a follow-up study using a similar approach, Detsky et al. confirmed the importance of older age, and supraventricular and ventricular arrhythmias, as independent predictors of perioperative cardiac risk.6 In a prospective evaluation, the Perioperative Ischemia Research Group identified left ventricular hypertrophy by ECG as one of five independent predictors of postoperative myocardial ischemia.7 Postoperative ischemia, in turn, was strongly correlated with the development of serious postoperative cardiac complications, including MI, heart failure, ventricular tachycardia, and cardiac death.8 More recently, Landesberg et al. found that left ventricular hypertrophy and ST-segment depression on ECG were the only two preoperative variables predictive of postoperative MI or cardiac death after major vascular surgery.9 Finally, in the largest prospective series reported to date, involving 4,315 patients aged 50 and older undergoing major elective noncardiac surgery, history of ischemic heart disease, as evidenced by Q-waves on the ECG or clinical history of MI, was one of six independent predictors of major cardiac complications.10 Although it is true, as indicated by Liu et al., that several other studies have failed to identify preoperative ECG abnormalities as independent predictors of cardiac risk,11–14 most of these studies were smaller in size or were limited to younger, healthier patients. Indeed, based on currently available evidence, the recently revised American College of Cardiology/American Heart Association Guideline on Perioperative Cardiovascular Evaluation for Noncardiac Surgery specifies “previous MI by history or pathologic Q-waves” as an intermediate-level indicator of increased risk of MI, heart failure, or death, whereas other ECG abnormalities, including left ventricular hypertrophy, nonsinus rhythm, left bundle branch block, and ST-T abnormalities are considered minor risk factors for postoperative cardiac events.15 The methodology used by Liu et al. is similar to that of earlier reports.4 These investigators evaluated 513 consecutive patients aged 70 and older undergoing major noncardiac surgical procedures. Preoperative data were collected by chart review, and all patients were followed prospectively for the development of perioperative and postoperative cardiac and noncardiac complications. Two investigators masked to subjects' clinical course classified preoperative ECG abnormalities according to Minnesota code criteria. Perioperative and postoperative complications, including ischemia, MI, arrhythmias, and heart failure, were identified using predefined criteria. The principal findings were that, although 75% of patients had at least one ECG abnormality and 10% of patients developed cardiac complications, in multivariate analysis preoperative ECG abnormalities were not associated with increased risk of procedure-related cardiac events. In this sample, American Society of Anesthesiologists physical status classification greater than 3 and history of heart failure were the only independent predictors of cardiac complications. The authors concluded that the preoperative ECG is “of limited value in predicting postoperative cardiac complications in older patients undergoing noncardiac surgery.” There are several important limitations to this study. First, the sample size is too small to reliably exclude the possibility that preoperative ECG abnormalities predict postoperative cardiac complications. This point is further underscored by the fact that the number of “hard” cardiac endpoints was very small (nine MIs and three cardiac deaths). Second, although the authors report that the presence of “any abnormality” on ECG did not predict postoperative cardiac events, Table 4 indicates that many nonspecific ECG abnormalities were included in this broad definition (e.g., left axis deviation (19.3% of the population) and minor T-wave changes (22.2% of the population)). It is thus possible that a more restrictive definition, such as the presence of major Q-waves, may have provided more useful prognostic information. Third, although the investigators who interpreted the ECGs were blinded to clinical events, it is not clear that event classification was conducted without access to the preoperative ECGs. Because criteria for diagnosing specific cardiac events are not defined in the manuscript and, in some cases, may be quite subjective (e.g., occurrence of chest pain), there is potential for introducing bias unless both the endpoints classification and ECG analyses were conducted in blinded fashion. Fourth, as with most single-center studies, the generalizability of the authors' findings to other patient populations at other institutions is unknown. Despite these limitations, the authors' conclusion that preoperative ECG abnormalities are less useful than other parameters, such as the presence of major comorbidities, active ischemia, or heart failure, in assessing perioperative cardiac risk in older adults undergoing major noncardiac surgery is probably correct. The general lack of sensitivity and specificity of the 12-lead ECG for diagnosing cardiac disease is well recognized, and, as the authors point out, prior studies addressing this issue have yielded mixed results.5–14 Moreover, although the cost of a single ECG is modest, the total annual cost of performing preoperative ECGs for every older adult undergoing major noncardiac surgery nationwide is substantial. Is it time, then, to dispense with the preoperative ECG altogether? Or, perhaps more conservatively, should we obtain preoperative ECGs only in patients deemed to be at moderate or high risk of cardiac events based on other clinical considerations? In my opinion, the answer to these questions is “no.” Apart from its role in predicting cardiac complications, there are at least two other major reasons for performing an ECG at the time of surgery or other hospitalization in older adults.16 First, the preoperative ECG provides a window into the diagnosis of unsuspected coronary heart disease, an important issue in older adults, because at least one-fourth of MIs in this population are “silent” or clinically unrecognized,17,18 and the risk of recurrent cardiac events in these patients is similar to that in individuals with recognized MIs.18,19 Second, as the authors suggest, the preoperative ECG provides an important “baseline” in the event that the patient develops chest pain or other cardiac complications postoperatively. In such cases, lack of a preoperative ECG for comparison may lead to misinterpretation of postoperative ECG abnormalities, potentially resulting in the performance of unnecessary tests and procedures, prolongation of hospitalization, and heightened risk for iatrogenesis. In our ongoing quest to provide high-quality and cost-effective care to all patients, regardless of age, it is entirely appropriate to periodically reassess the value of “standard practices” to ensure that such practices remain concordant with evolving concepts of quality, effectiveness, and cost. In this regard, Liu et al. have provided an important contribution to the decades-old debate about the clinical utility of the routine preoperative ECG, but, rather than fostering what, in my view, would be a premature move to cut back on the use of preoperative ECG testing, it is hoped that the present study will fuel further research to better define specific situations for which the preoperative ECG may truly be superfluous. This might include, for example, low-risk patients undergoing minor surgical procedures, for whom perioperative β-blocker therapy may provide sufficient prophylaxis against the development of cardiac complications.20 In the meantime, it is my view that the preoperative ECG frequently provides valuable information that facilitates perioperative management, and I will continue to request ECGs for all my older patients undergoing major noncardiac surgical procedures.