From the Divisions of Cardiology (Drs Hernandez, Newby, and O'Connor) and Clinical Pharmacology (Dr O'Connor), Department of Medicine, Duke University Medical Center and Duke Clinical Research Institute, Durham, NC. Dr Newby has received research grant support from Roche Diagnostics Corporation and Dade-Behring Inc, participates in a speaker's bureau sponsored by Biosite, and serves as a consultant for Ortho Clinical Diagnostics.
The number of patients undergoing major noncardiac surgery has steadily increased over the last decade. Cardiovascular complications are important and often feared by patients, surgeons, and anesthesiologists. Although preoperative risk assessment has improved since Goldman and colleagues published their landmark article that introduced the Multifactorial Index of Cardiac Risk 25 years ago, it continues to require modification, especially with the increasing prevalence of heart failure and the increase in procedures performed in the elderly. This review will summarize preoperative assessment and perioperative management with an emphasis on heart failure.
Over the last 25 years, there has been steady improvement in the care of patients undergoing major noncardiac surgery. In the past, attention was focused on coronary artery disease (CAD) and the detection of ischemia. For the future, changes in the epidemiologic characteristics of patients undergoing surgery may require improved strategies and care, especially for those with heart failure (HF). This review will summarize preoperative risk evaluation and perioperative care focusing on HF.
The increasing volume of noncardiac surgical procedures and changes in the epidemiology of cardiovascular disease may collide to create an epidemic of postoperative complications despite improved care in both fields (Figure 1). In 2000, there were almost 40 million procedures performed in the United States, with over 10 million major noncardiac surgical procedures compared with only 519 000 cardiac surgical procedures.2 The elderly pose a special problem since they are the largest population to undergo surgery, yet are at high risk for perioperative complications and major cardiac complications.3,4 Individuals 65 years and older account for over 10 million noncardiac procedures and at least 4 million major noncardiac surgical procedures each year, which has increased over the last decade.2
The noncardiac surgical volume is increasing beyond previous predictions by Mangano1 in 1990.
Based on information from the late 1980s, at least 1 million perioperative cardiac complications occur per year, with an estimated $20 billion in annual costs for in-hospital and long-term care.1 Although studies show clinical pathways can reduce length of stay, there may be vulnerable groups, such as the older, poorer, or sicker populations, which may be more susceptible to adverse outcomes with subsequent costs occurring after the initial discharge.5- 7 Currently, no accurate estimates are available owing to improvements in anesthesia techniques, β-blocker use, and risk assessment vs an increased prevalence of HF, an aging population, and more survivors of sudden death and myocardial infarction (MI) (Figure 2).
Despite advances in care, the changing complexity of patients undergoing major noncardiac surgery may make prediction of cardiovascular risk more difficult in the future. CHF indicates congestive heart failure; MI, myocardial infarction.
The incidence and prevalence of HF is soaring at a staggering rate. Based on the Framingham Heart Study, there are an estimated 550 000 new cases of HF each year, with an estimated prevalence of 5 million patients.8
Several factors may contribute to the increasing incidence and prevalence of HF. Advances in medical care, in particular strategies for management of acute ischemic heart disease and sudden death, have resulted in improved survival with chronic diseases such as atherosclerosis, hypertension, and diabetes.9,10 Recently, the Framingham Heart Study also showed an improvement in survival of patients diagnosed with HF, which, coupled with a rapidly growing elderly population, could also contribute to a higher long-term prevalence.11,12
These changes in the epidemiology of cardiovascular disease along with an aging population will change substantially the landscape of preoperative risk stratification and raise concerns whether our current strategies are adequate.
Goldman and colleagues13 established a multivariable model of cardiac risk for noncardiac surgery, later named the original Cardiac Risk Index, which dramatically improved the prior American Society of Anesthesiologists (ASA) classification.14 There have been other models developed to assess risk for perioperative complications emphasizing different aspects. All use clinical assessment, medical history, and a few basic laboratory assessments and now serve as a routine component of preoperative screening.
In their original study, Goldman and colleagues13 enrolled 1001 patients 40 years or older undergoing major noncardiac surgery. They derived a multivariable model using 9 clinical signs and standard laboratory measurements to generate a weighted cardiac index with 4 classifications predicting cardiac risk independent of surgery.13 This served as the basis for many years for most physicians assessing preoperative risk. However, limitations such as lack of model validation, unknown interobserver variability, and expectation bias due to event classification by investigators led to the development of other risk indices.
The next major development arose from a study of 455 consecutive patients referred to a general medical consultation service.15 Detsky and colleagues15 found an area under the receiver operator characteristic (ROC) curve of 0.69 for the original Cardiac Risk Index, which was lower than that in the original study by Goldman and colleagues.13 They simplified the point system, added angina severity, and modified the criteria for HF, providing an improved c-index of 0.76. While the efforts of Detsky and colleagues15 were a substantial improvement, there were concerns of referral bias given that surgeons or anesthesiologists had to ask for a medical consultation. Overall, the main limitations were the single institution design, lack of end point adjudication, and the limits of generalizability of the risk index to "real-world" consultation.
In an effort to further improve available risk stratification tests, Lee and colleagues16 proposed an even easier risk stratification tool composed of 6 simple factors derived from the largest study cohort to date—4315 patients 50 years or older identified through the hospital's Preadmission Test Center or in the hospital (Table 1) and (Table 2). Through this mass screening process, approximately 80% of patients undergoing major noncardiac surgery were approached for enrollment into the study. Major cardiac complications were seen in 2% of patients in the derivation cohort and 2.5% in the validation cohort. Independent predictors of risk in this cohort were high-risk type of surgery (intraperitoneal, intrathoracic, or suprainguinal vascular), history of ischemic heart disease, history of HF, history of cerebrovascular disease, preoperative treatment with insulin, and preoperative serum creatinine level greater than 2.0 mg/dL (>176.8 µmol/L). The area under the ROC curve for this model was 0.806 in the validation cohort, compared with 0.582 for the Modified Cardiac Risk Index, 0.701 for the original Cardiac Risk Index, and 0.706 for ASA class.
Overall, this simplified approach effectively grouped patients into 3 groups. What to do for low-risk patients is straightforward, but those classified as intermediate or high risk still require clinical judgment for the best approach. The investigators used an improved study design by having study personnel who did not participate in subjects' care perform daily medical review to collect data, and a reviewer blinded to preoperative clinical data classified all postoperative outcomes. Of note, possibly related to the smaller size of the validation cohort, was that insulin therapy and preoperative serum creatinine level did not prove to be as important in the validation phase. Another limitation is that the model cannot take into account changes in a patient's clinical status over time. For example, if a patient has decompensated HF today, the physician delays surgery for 1 month, and the patient clinically improves, the patient's calculated risk remains the same, which may or may not reflect reality. The same applies to a patient with a recent acute coronary syndrome who has surgery delayed for months after coronary revascularization.
In the original Cardiac Risk Index by Goldman and colleagues,13 clinical signs of HF including an S3 gallop or jugular venous distention (JVD) were the most significant predictors of postoperative life-threatening or fatal cardiac complications. In the final analysis, signs of HF carried the highest weight in the original Cardiac Risk Index. In addition, 36 of the 39 patients manifesting 1 or more life-threatening cardiac complications had pulmonary edema (Table 3).
Owing to the diverse group of clinicians evaluating numerous patients, in actual practice, the prior assessment for HF soon was recognized to be impractical. In the study by Detsky and colleagues,15 the interobserver agreement for S3 and JVD was poor (κ statistic, 0.42 and 0.50, respectively). Therefore, to make the diagnosis of HF more objective and reproducible preoperatively, Detsky and colleagues grouped HF into 2 categories as the presence of alveolar pulmonary edema within 1 week or ever. Although the definition was stricter, HF still had a major role in predicting events as well as being a major outcome. Of the 43 serious events, there were 10 new or worsened episodes of HF without alveolar pulmonary edema, and 5 episodes of alveolar pulmonary edema.15
In the Revised Cardiac Risk Index study population, HF was both an important predictor and a key complication. The outcome required a formal reading of pulmonary edema on the chest radiograph by a radiologist with a plausible clinical setting. In the validation set, it provided the highest odds ratio (4.3) for major cardiac complications. In addition, it was an important outcome similar to MIs.16 Other models have shown HF as an important predictor of perioperative events and the degree of HF appears to correlate with complications.17
Despite use of the standard Framingham criteria for HF in a number of large epidemiological studies, investigators have not clearly applied these same standard definitions in preoperative risk studies.18,19Table 4 summarizes the definitions of HF that have been used in the major preoperative risk studies.
Preoperative assessment of HF has shifted from emphasizing physical examination signs to simply having a history of HF. In doing so, problems of interobserver variability and reproducibility in general application were reduced. In the Revised Cardiac Risk Index, the investigators defined HF by a combination of symptoms and signs that incorporated some of the Framingham criteria, but important elements were still missing such as orthopnea and dyspnea on exertion that are now recommended by the American College of Cardiology/American Heart Association (ACC/AHA) guidelines for the definition of HF.16,18,20
The studies required a postoperative chest radiograph for confirmation of HF, which helped standardize the diagnosis.13,15,16 However, all of the studies depended on the primary provider to order chest radiographs since it was not done routinely in the postoperative period, causing end points to be missed. For most of the studies, a single reviewer classified events rather than a central events classification committee.13,15- 17,21,22
Over the last several years, numerous studies have highlighted the importance of natriuretic peptides as diagnostic and prognostic markers in HF.23- 26 With the recent approval of commercial assays for B-type natriuretic peptide and N-terminal pro–B-type natriuretic peptide, it may be possible to improve both preoperative classification of HF and diagnosis of HF as a postoperative complication by incorporating the markers in routine assessment. Further studies will be needed to assess the utility of such a strategy.
Although echocardiography appears to be very accessible and potentially useful test for preoperative evaluation, it has limited prognostic value as a routine test. To help determine the value of routine echocardiography in preoperative screening, Halm et al27 evaluated 339 of 474 consecutive men in a Veterans Affairs medical center with known or suspected CAD. In this study, 8% of patients had HF and 3% had ischemic events postoperatively. Although an ejection fraction less than 40% and wall motion score were associated with some outcomes, both had poor predictor characteristics.
In a more recent study, Rohde and colleauges28 evaluated 570 patients enrolled in the Revised Cardiac Risk Index cohort who underwent transthoracic echocardiography at the discretion of their physician within 3 months of surgery. Overall, models including the reported echocardiographic data predicted major cardiac complications better than models with only clinical variables (c-statistic, 0.73 vs 0.68; P<.05). An abnormal echocardiogram with any degree of systolic dysfunction, moderate to severe left ventricle (LV) hypertrophy, moderate to severe mitral regurgitation, or aortic gradient of 20 mm Hg or greater provided a sensitivity of 80%, specificity of 52%, positive predictive value of 12%, and negative predictive value of 97%. However, severe LV dysfunction compared with mild to moderate LV dysfunction did not have as strong an association with cardiogenic pulmonary edema and MI. The heterogeneity of these findings likely points out that HF and ischemic heart disease comprise a combination of factors that change every day, while we usually only measure a few at 1 point in time. Therefore, in the end, adding echocardiography added little to risk models.
A number of studies have evaluated stress imaging in preoperative risk assessment; the largest experience using dipyridamole thallium-201 imaging.29 Eagle and colleagues21 used a combination of clinical and thallium data for preoperative assessment in patients with vascular disease. The latest study from his group used a model incorporating the clinical variables of advanced age (>70 years), angina, history of MI, diabetes mellitus, history of HF, and prior coronary revascularization.22 Importantly, even in that study that emphasized ischemic heart disease, HF was an important predictor of the primary outcome (nonfatal MI, fatal MI, or cardiac death) in the training set and validation set with odds ratios of 2.7 and 3.2, respectively. Although thallium scoring correlated with events, the results showed that only the intermediate-risk group based on clinical assessment benefited from further testing to stratify patients into low- and high-risk categories.
As observed by Eagle and colleagues,21 HF played a major role as a clinical predictor in other studies evaluating perfusion imaging. In a meta-analysis combining 5 studies using thallium imaging for preoperative risk assessment in 1188 patients, HF was the second most important predictor of cardiac events behind reversible thallium defect (odds ratio, 3.6; P<.001).29
With today's current approach to preoperative care, it is unclear what the risk factors for postoperative HF are. In a 1990 study by Mangano and colleagues,30 a history of dysrhythmia, diabetes, duration of anesthesia, vascular surgery, and narcotic anesthesia were all associated with postoperative HF while postoperative ischemia was not. They speculated that vascular surgery placed patients at higher risk because of the length of the procedure and volume of intravenous fluids. In another study, Charlson and colleagues31 found that the risk for postoperative HF was limited to patients with preoperative symptomatic cardiac disease, especially in patients with diabetes. Another possible factor in postoperative HF may be the inability to administer some HF medications because of the inability to use an oral route postoperatively. Finally, it appears that surgery leads to activation of the renin-angiotensin system and postoperative elevations in cortisol as well as epinephrine levels.32
Manifestations of perioperative HF usually develop during the day of surgery or the second to third postoperative day, although this is not well studied.31 Theoretically, there are 2 potential periods when HF may worsen: (1) immediately after surgery because of the length of surgery, myocardial ischemia, and rapid fluid shifts and (2) a few days later, when HF may occur because of reabsorption of third-spaced fluid.33 In the study by Mangano and colleaugues,30 48% of HF occurred after the third postoperative day.
The exact approach to patients with HF in the perioperative period is uncertain, but understanding the degree and cause may be helpful. The ACC/AHA guidelines state, "Every effort must be made to detect unsuspected HF by a careful history and physical examination."34
The mainstay of perioperative care is to identify patients with intermediate to high risk features and either perform preoperative evaluation with noninvasive testing followed by revascularization or use perioperative β-blockers (Table 5). There are no randomized controlled trials indicating the effect of revascularization preoperatively, and previous observational studies provide a mixed picture.35- 40 Until a large randomized trial is done, the benefit of revascularization remains to be definitively proven for prevention of perioperative cardiac events.
Randomized controlled trials support β-blocker therapy perioperatively, but there have only been 30 patients with HF in these trials.41- 45 The first major trial using atenolol in 200 patients with CAD or at risk for CAD showed a reduction in deaths and combined cardiovascular outcomes.45 However, the benefit was not apparent until after discharge and was largely seen during the first 6 to 8 months after surgery, although the protocol only stipulated therapy during hospitalization.
The Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography (DECREASE) study also supports the merits of perioperative β-blocker therapy.44 Of the 846 patients screened with dobutamine echocardiography, 112 with positive test results were ultimately randomized to perioperative treatment with bisoprolol or standard care. The primary end point of death from cardiac causes or nonfatal MI was significantly different between the randomized groups; 2 patients (3.4%) in the bisoprolol group compared with 18 patients (34%) in the standard care group (P<.001) experienced the primary end point. Of note, 8 patients with extensive wall-motion abnormalities either at rest or during stress testing were excluded. If the Revised Clinical Risk Index is considered, dobutamine stress echocardiography only adds significant prognostic information for patients with 3 or more risk factors.46 Thus, in most patients who have fewer than 3 risk factors, use of perioperative β-blockade without having to undergo substantial testing appears reasonable.47 However, the study was limited in describing an approach for patients with HF since it only had 14 patients with HF and HF was not an outcome.
One issue that arises in surgical patients that may be particularly important for patients with HF is how to administer their medications appropriately if patients are not able to take anything orally. Clinicians use intravenous medications such as β-blockers in the perioperative period, but timing or duration is uncertain in patients with worsening HF, given the small number of patients in clinical trials. Moreover, the pharmacodynamics of intravenous medications may not be as well appreciated by clinicians compared with the more routinely used oral agents.
For patients with HF, some investigators recommend right heart catheterization (RHC) in the perioperative period depending on the clinical situation.48,49 Intraoperative hemodynamic changes are associated with increased complication rates, so logically most would assume invasive monitoring would reduce perioperative complications.50
In an observational cohort study of 4059 patients who underwent major noncardiac surgery (excluding abdominal aortic aneurysm repair), the value of RHC was underwhelming. Over 200 patients had an RHC, with an overall 3-fold increase in the incidence of major postoperative cardiac events and an adjusted odds ratio of 2.0 for postoperative major cardiac events.51
In the most recent study evaluating RHC, investigators showed no benefit for perioperative RHC in a randomized controlled trial of elderly patients undergoing major noncardiac surgery.52 Almost 2000 patients 60 years or older with ASA's class III-IV scheduled for major surgery followed by a stay in an intensive care unit were randomized to RHC or usual care. There was no benefit to RHC-directed therapy over standard care, and there was a higher rate of pulmonary embolism in the catheter group. One limitation of this study is that New York Heart Association (NYHA) class III-IV HF patients only composed 13% of the study population. Thus, it is unknown whether RHC helps in this group, but the trend favored standard care in the subgroup analysis. The study provided hemodynamic goals for the RHC group resulting in higher use of inotropic agents (48.9% vs 32.8% in the standard care group), which may be a reason for the lack of benefit. Thus, we see another area in which clinical logic does not correlate with clinical outcome.
Transesophageal echocardiography (TEE) and continuous electrocardiogram evaluation have also been evaluated to determine if routine use in patients at high risk for CAD adds to the identification of perioperative outcomes beyond clinical assessment. However, sensitivity and positive predictive value were low for TEE prediction of ischemic outcomes. In a multivariable analysis, the odds ratio for the association of TEE findings with outcomes was 2.6 and was most predictive of ventricular tachycardia. In evaluating ischemic outcomes alone, TEE was not predictive beyond routine clinical data.53
Preoperative evaluation has progressed substantially, but several limitations could be addressed in the future. Most of the prior studies were done in a single center, and only recently have studies incorporated multiple centers.54,55 However, these ongoing studies may be limited owing to the unique population within the Veterans Affairs system or because of the difficulty in generalizing from academic centers to community practice. Future studies must also address prior study limitations such as end point adjudication with a multiple reviewer system and standardized evaluation of HF outcomes. Finally, assessing early events after discharge and readmissions that may be related to perioperative cardiovascular events should be captured.
Now it is time to turn to evaluation of other medications and other preoperative risk stratification strategies, especially in patients with HF. Physicians do not routinely follow evidenced-based guidelines for risk stratification and testing strategies, while at the same time their estimates of cardiac risk compared with validated indices are poor.56 Thus, the future will require a diverse group of clinicians to efficiently and accurately evaluate a large number of patients possibly at higher risk.
Medications, such as statins, which were shown in a recent case-control study and another observational study to lower risk of perioperative cardiac complications, should be validated prospectively.57,58 In patients with HF or at risk for HF when undergoing noncardiac surgery, studies should evaluate the role of digoxin and angiotensin-converting enzyme inhibitors as well as other new therapies.
Measurement of multiple biomarkers such as natriuretic peptides or C-reactive peptide may provide a simple and efficient method of risk stratification compared with undergoing expensive noninvasive studies or even cardiac catheterization. In patients with NYHA class III-IV HF, the role of RHC or other noninvasive tests such as bioimpedance before surgery may be useful in addition to new therapies that may be available without as many adverse effects as currently used inotropic agents.
Finally, because of the pressure to reduce length of stay and prevent readmission, studies should evaluate new strategies, such as measurements of B-type natriuretic peptide levels, to aid decision making for discharge of patients with HF. Thus, there are many potential areas of investigation needed to reduce the morbidity or mortality of patients with HF undergoing major noncardiac surgery.
Patients with HF have as significant a risk of perioperative complications as patients with ischemic heart disease, and they should be evaluated thoroughly to ensure that they are well compensated prior to undergoing surgery and that standard perioperative β-blockade is used. Excessive testing with echocardiography or RHC does not necessarily assist in determining safety for surgery. Stress testing for high-risk patients may help in deciding proper therapies or revascularization strategies, while those at intermediate risk may only need β-blocker therapy. In most well-planned elective surgical procedures, the issues of stability while receiving β-blockers should already be settled. Postoperative monitoring for decompensation, appropriate medication use, and early follow-up by a primary care provider should help further reduce perioperative complications.
There remain a number of unresolved issues, and perioperative care may need to be improved in patients with complex cardiovascular disease such as HF. Last minute cardiac consultations are already a part of daily care in the preoperative setting.59 Therefore, pathways for risk assessment and intervention should be simplified. Prior studies have emphasized ischemic heart disease, but with emerging techniques such as the measurement of biomarkers, more accurate determination of HF in the perioperative setting may lead to further refinements in prognosis or diagnosis in patients undergoing noncardiac surgery.
Correspondence: Christopher M. O'Connor, MD, DUMC Box 3356, Duke University Medical Center, Durham, NC 27710 (firstname.lastname@example.org).
Accepted for publication October 10, 2003.
Dr Hernandez is supported in part by an American College Cardiology Foundation/Merck grant.
Thank you for submitting a comment on this article. It will be reviewed by JAMA Internal Medicine editors. You will be notified when your comment has been published. Comments should not exceed 500 words of text and 10 references.
Do not submit personal medical questions or information that could identify a specific patient, questions about a particular case, or general inquiries to an author. Only content that has not been published, posted, or submitted elsewhere should be submitted. By submitting this Comment, you and any coauthors transfer copyright to the journal if your Comment is posted.
* = Required Field
Disclosure of Any Conflicts of Interest*
Indicate all relevant conflicts of interest of each author below, including all relevant financial interests, activities, and relationships within the past 3 years including, but not limited to, employment, affiliation, grants or funding, consultancies, honoraria or payment, speakers’ bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued. If all authors have none, check "No potential conflicts or relevant financial interests" in the box below. Please also indicate any funding received in support of this work. The information will be posted with your response.
Some tools below are only available to our subscribers or users with an online account.
Download citation file:
Web of Science® Times Cited: 13
Customize your page view by dragging & repositioning the boxes below.
Enter your username and email address. We'll send you a link to reset your password.
Enter your username and email address. We'll send instructions on how to reset your password to the email address we have on record.
Athens and Shibboleth are access management services that provide single sign-on to protected resources. They replace the multiple user names and passwords necessary to access subscription-based content with a single user name and password that can be entered once per session. It operates independently of a user's location or IP address. If your institution uses Athens or Shibboleth authentication, please contact your site administrator to receive your user name and password.