Reevaluating the Cardiac Risk of Noncardiac Surgery Using the National Surgical Quality Improvement Program

Recommendations from the appropriate use criteria for the evaluation of stable ischemic heart disease suggest that preoperative cardiac stress testing is rarely appropriate when used in the context of low-risk noncardiac surgery. Guidelines for the preoperative evaluation of patients prior to noncardiac surgery recommend against routine preoperative cardiac testing in patients at low cardiac risk, as indicated by a 30-day postoperative incidence of death or myocardial infarction <1%. Previous guidelines from the American College of Cardiology and American Heart Association, as well as current guidelines from the European Society of Cardiology (ESC) and European Society of Anaesthesiology (ESA), have described low-, intermediate-, and high-risk categories of noncardiac surgery based on the expected incidence of postoperative myocardial infarction and cardiac arrest. As surgical techniques have evolved and patient outcomes improve over time, however, it is unclear whether the categorizations of cardiac risk described in these guidelines remain accurate. The goal of our study was to reevaluate and newly categorize the cardiac risk of elective, nonemergent, noncardiac surgeries using patient data and outcomes from the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) database.

We reviewed data from the 2018 ACS NSQIP Participant Use Data File (PUF). This database includes patient-level surgery data and 30-day surgical outcomes collected prospectively from 722 participating sites throughout the United States. We reviewed the most recently available ACS NSQIP PUF, which includes data for surgeries performed at participating sites from January through December 31, 2018. We excluded data from cardiac surgeries and from surgeries that were designated in the PUF as either emergent or nonelective. Surgeries were then sorted into broad categories including: abdominal, endocrine, gynecologic, head/neck, orthopedic, spinal, superficial, thoracic, urologic, vascular, and other, similar to the approach described by Glance et al in developing their surgical mortality risk model. Within each category, we grouped Current Procedural Terminology (CPT) codes based on anatomical position, procedure type, and surgical approach, consistent with CPT surgical groupings defined by the American Academy of Professional Code. We extracted patient clinical risk factors for coronary artery disease including hypertension, diabetes mellitus, and smoking status. We also extracted data about preoperative creatinine, patient functional status (independent vs dependent), and American Society of Anesthesiologists physical status classification, which variables were previously identified as factors associated with postoperative myocardial infarction and cardiac arrest. We also recorded patient age and sex, as well as preoperative white blood cell count and hematocrit. For patients in which age was recorded only as “90+” in the NSQIP PUF, an age of 90 years was assigned for analysis purposes.

We reviewed the PUF for incidence of postoperative major adverse cardiac events (MACE). We defined postoperative MACE as all-cause mortality, myocardial infarction, or cardiac arrest occurring within 30 days of surgery to reflect definitions of MACE and cardiac risk used in the American College of Cardiology/American Heart Association and the ESC/ESA perioperative guidelines. For each surgical grouping, we used the Clopper-Pearson method to determine 95% confidence intervals (CI) for MACE. According to the 95% CI for postoperative MACE, surgical groups were categorized as low risk (95% CI <1%), intermediate risk (95% CI above and below 1%), or elevated risk (95% CI ≥1%).

Multivariable adjusted logistic regression analyses were performed for the purpose of determining the odds of postoperative MACE, and the odds of the individual components thereof, for the intermediate and elevated-risk categories relative to the low-risk category, while simultaneously controlling for various potentially confounding demographic and clinical risk factors. Surgical cases with missing variable data were excluded. Age, preoperative creatinine, preoperative hematocrit, and preoperative white blood cell count were evaluated as continuous variables in the logistic regression model. Reference values for the remaining categorical variables in the logistic regression analyses were assigned as follows: sex (female), hypertension (none), diabetes mellitus (none), smoking (none), functional class (independent), American Society of Anesthesiologists physical status class (≤2). SPSS was used for the data analysis (IBM SPSS Statistics for Windows, Version 27.0, IBM Corp, Armonk, NY). This study was submitted to the Institutional Review Board at the Penn State College of Medicine and determined to be exempt from review.

The 2018 PUF data file included a total of 1,020,511 surgical cases, of which 807,413 cases were included in the analysis after excluding nonelective surgeries, emergent surgeries, and cardiac surgeries. Postoperative MACE were reported in 4047 of 807,413 patients (0.50%) undergoing elective, nonemergent noncardiac surgery. Patients with postoperative MACE were more likely to be male, to be older, and to have higher creatinine, higher white blood cell count, and lower hematocrit compared with those without postoperative MACE. Patients with postoperative MACE were also more likely to smoke and to have diagnoses of hypertension and diabetes mellitus compared with patients without postoperative MACE. A detailed comparison of patient characteristics among those who did and did not experience postoperative MACE is shown in Table 1.

Eighty-seven groups of surgery were defined based on surgical category, anatomical location, and surgical approach. Of these, 40 were low risk (comprising 667,735 procedures), 29 were intermediate risk (comprising 53,499 procedures), and 18 were elevated risk (comprising 86,179 procedures). Postoperative MACE were reported in 1708/667,735 (0.26%) of the low-risk category, in 516/53,499 (0.96%) of the intermediate-risk category, and in 1823/86,179 (2.12%) of the elevated-risk category. The elevated-risk category accounted for 10.7% of total procedures, and 45.1% of total postoperative MACE. Table 2, Table 3, and Table 4 show the incidence of postoperative MACE for the low-risk, intermediate-risk, and elevated-cardiac-risk categories, respectively.

When compared with the low-risk category, the intermediate-risk category had increased multivariable adjusted risk for postoperative MACE (adjusted odds ratio [aOR] 2.35; 95% CI, 2.12-2.62), myocardial infarction (aOR 1.76; 95% CI, 1.48-2.09), cardiac arrest (aOR 2.42; 95% CI, 1.97-2.98), and all-cause mortality (aOR 2.98; 95% CI, 2.59-3.43). The elevated-risk category, when compared with the low-risk category, had increased multivariable adjusted risk for postoperative MACE (aOR 3.15; 95% CI, 2.92-3.39), myocardial infarction (aOR 2.71; 95% CI, 2.42-3.02), cardiac arrest (aOR 3.64; 95% CI, 3.15-4.21), and all-cause mortality (aOR 3.59; 95% CI, 3.23-3.98).

An accurate understanding of the cardiac risk associated with noncardiac surgery is essential to inform individual and institutional protocols about the appropriate use of preoperative cardiac evaluation and testing. Using the NSQIP database, we stratified noncardiac surgeries into low-, intermediate-, and elevated-risk categories based on their incidence and 95% CI for postoperative MACE, which categories may be useful in identifying populations most likely to benefit from preoperative cardiac testing.

The need to reevaluate the cardiac risk of noncardiac surgeries is essential as patient outcomes continue to improve over time. Using the Nationwide Inpatient Sample, Semel et al found a significant decrease in 30-day postoperative mortality following cardiac and noncardiac surgeries over time (1.68% in 1996 to 1.32% in 2006). Smilowitz et al  analyzed 10,581,621 noncardiac surgeries over a 10-year period from 2004 to 2013 and similarly found improved outcomes, including a 0.5% reduction (3.1% to 2.6%) in major adverse cardiovascular and cerebrovascular events. Campos et al found that, as bariatric surgery evolved from an entirely open approach in 1993 to a primarily laparascopic approach in 2016, patient mortality (0.6% to 0.04%) and adverse cardiovascular and thromboembolic events (0.9% to 0.3%) significantly improved over that time period. Gupta et al, in their study developing the NSQIP Myocardial Infarction and Cardiac Arrest risk calculator, found postoperative myocardial infarction and cardiac arrest to occur in 0.65% of patients in the NSQIP PUF from 2007, as compared with a 0.33% incidence of postoperative myocardial infarction and cardiac arrest in this study of the NSQIP PUF from 2018. With improving outcomes over time, cardiac risk models likely require periodic re-calibration and validation with newer datasets to ensure continued accurate discrimination of perioperative cardiac outcomes.

We found similarities as well as important distinctions in our data, compared with existing models for categorizing the cardiac risk of noncardiac surgeries. For example, the Revised Cardiac Risk Index (RCRI) defines high-cardiac-risk surgery as intraperitoneal, intrathoracic, and suprainguinal vascular surgery. We similarly found elevated-risk surgeries to be associated with the abdominal, thoracic, and vascular categories, in agreement with the RCRI. However, multiple abdominal/intraperitoneal procedures were also associated with a low risk of postoperative MACE, particularly in the setting of a laparoscopic approach, which otherwise would be inappropriately assigned as high risk using the RCRI model. Whether an adjustment to the assignment of high-risk surgery in the RCRI based on this recent data could improve discrimination of cardiac outcomes by the RCRI is worth further investigation.

The current ESC/ESA perioperative guidelines indicate that major orthopedic surgeries such as hip surgery and spine surgery, as well as major urological and gynecological procedures, are associated with postoperative MACE in the range of 1%-5%. In contrast to this, with exceptions such as kidney and bladder surgery, as well as hip surgery excluding total hip arthroplasty, we found that procedures in the categories of orthopedic, spinal, urologic, and gynecologic surgery were most commonly associated with low risk (<1%) of postoperative MACE. Even surgeries with elevated postoperative MACE in our study, such as open esophageal surgery (3.6%; 95% CI, 2.7%-4.7%) and pneumonectomy (2.9%; 95% CI, 2.1%-3.8%) had lower adverse cardiac events than suggested in the ESC/ESA guidelines, which predict a postoperative MACE >5% for both of these procedures. These discrepancies may result from improvement in surgical outcomes over time, as previously discussed. Additionally, data for the cardiac-risk categories in the ESC/ESA guidelines were derived from a study of postoperative mortality and not of postoperative cardiac events, which may have resulted in an inaccurate categorization of cardiac risk as described in the guidelines.

Updated estimation of surgical risk categories will help clinicians to appropriately apply preoperative societal recommendations and appropriate use criteria in the setting of patients undergoing low-risk noncardiac surgery. For example, the Multimodality Appropriate Use Criteria for the Detection and Risk Assessment of Stable Ischemic Heart Disease provide specific criteria for the use of preoperative cardiac testing in patients undergoing low-risk or intermediate-risk noncardiac surgery, without specifying what surgeries are associated with low cardiac risk or intermediate cardiac risk. Additionally, societal guidelines in the Choosing Wisely campaign describe multiple scenarios in which preoperative cardiac testing is inappropriate prior to low-cardiac-risk procedures, without specifying which procedures should be considered low cardiac risk. Categorizations described in this study may help to clarify which noncardiac surgeries are associated with low risk of postoperative MACE to avoid overuse of preoperative cardiac testing.

This study has a number of strengths, including incorporation of patient data and 30-day surgical outcomes collected from hundreds of institutions participating in the NSQIP database. Our study provides an important update in reporting categories of cardiac surgical risk compared wiith prior categorizations in major guidelines, which relied on either expert opinion or older surgical data. The surgical categories described in this manuscript include a broad range of procedures across multiple surgical specialties, and include distinctions in outcomes between open and laparoscopic surgical approaches.

Our study has a number of important limitations. For example, while many surgical procedures had thousands of data points in the NSQIP PUF, other less common procedures had significantly fewer data points, affecting the precision to which we could estimate cardiac risk in these procedures. Our study focused on elective, nonemergent procedures, and our results cannot be extrapolated to nonelective or emergent surgeries, which are likely to have relatively increased complications, including higher incidence of postoperative MACE. Postoperative troponin values are not systematically reported or collected in NSQIP, and therefore it is possible that postoperative myocardial infarction may be underestimated. Our study focused on postoperative myocardial infarction and cardiac arrest, while additional cardiovascular outcomes such as postoperative atrial fibrillation or congestive heart failure are not tracked in the NSQIP database and therefore were not included.

Beyond the surgical categories described here, factors such as patient age and comorbidities are associated with cardiac risk, and should be accounted for when determining individual patient risk. We demonstrated this in our study by showing significant differences in patient demographics and comorbidities between patients with and without MACE following noncardiac surgery, with many of these factors associated with increased risk of MACE on logistic regression analyses. NSQIP-based risk calculators such as the NSQIP Myocardial Infarction and Cardiac Arrest risk calculator and the Universal ACS Surgical Risk Calculator help to account for these differences by incorporating both patient and surgical risk factors to discriminate cardiac outcomes; however, as previously discussed, both models may be subject to overestimation of cardiac risk as patient outcomes have improved over time.

Periodic re-evaluation of patient outcomes using current surgical data is essential to provide an accurate understanding of the cardiac risk of elective, nonemergent noncardiac surgeries. In this study, we determined categories of low, intermediate, and elevated cardiac risk from the NSQIP database, which will help clinicians to guide preoperative cardiac testing and evaluation to patient populations who are most at risk for postoperative adverse cardiac events.

Fuente: The American Journal of Medicine, Vol 134, No 12, December 2021.   |  DOI:https://doi.org/10.1016/j.amjmed.2021.07.016