Abstract
Background:
Perioperative cardiovascular events are a leading cause of morbidity and mortality after non-cardiac surgery. We propose a simplified method for perioperative risk stratification.
Methods:
A retrospective cohort study identified patients undergoing non-cardiac surgery between 2009–2015 in the United States National Surgical Quality Improvement Program. Multivariable logistic regression models adjusted for age, sex, race and surgery type were generated to estimate the impact of traditional cardiovascular risk factors (hypertension, diabetes mellitus, current smoking) on odds of perioperative myocardial infarction (MI). Time to event analysis was conducted using competing risk analysis, with MI as the outcome event and death as the competing risk.
Results:
A total of 3,848,501 non-cardiac surgeries were identified. Post-operative MI occurred in 0.37% of patients and 1.04% of patients died. The 30-day event rate of perioperative MI increased in a stepwise fashion with additional risk factors (0.41% for one, 0.81% for two, and 1.07% for three; P-for-trend < 0.001) after accounting for the competing risk of death. In comparison to those with no risk factors, patients with one, two and three risk factors had increased odds of MI (aOR 2.07; 95% CI 1.96–2.19; aOR 3.63 95% CI 3.43–3.85; aOR 5.54 95% CI 5.09–6.04). Perioperative MI was rare (0.10%) in patients without risk factors.
Conclusions:
Patients with cardiovascular risk factors are at increased risk of perioperative MI, those without risk factors are at low risk. Further evaluation is needed to determine the impact of a simplified risk score in the perioperative setting.
Keywords: Cardiovascular Risk Factors, Perioperative Myocardial Infarction, Perioperative Mortality
Brief Summary:
The traditional cardiovascular risk factors of diabetes, hypertension and current smoking, alone and in combination, are associated with increased risk of perioperative myocardial infarction. In patients undergoing non-cardiac surgery, absence of these three risk factors is associated with a low risk of perioperative myocardial infarction. Clinicians may use this simplified risk stratification method to identify low risk surgical patients in the absence of laboratory values.
Introduction:
Myocardial infarction (MI) is a common cardiovascular complication of non-cardiac surgery and is independently associated with post-operative mortality.1 The prevalence of cardiovascular risk factors in patients undergoing non-cardiac surgery has increased over the past decade.2 To estimate risk of perioperative MI, risk scores have been developed from large cohorts.3,4 Unfortunately, modern risk scores frequently rely on the results of laboratory testing that may be unavailable at the time of initial clinical examination,1,3,5,6 rely on subjective assessments of risk,3 or are complex and cannot be easily calculated at the bedside.6 Furthermore, the impact of traditional cardiovascular risk factors on adverse outcomes after non-cardiac surgery has not been fully explored in contemporary practice.7,8 The absence of traditional cardiovascular risk factors may be sufficient to identify patients at a low risk for complications who do not require additional pre-operative testing or risk stratification. We evaluated the association between 3 traditional cardiovascular risk factors (hypertension, diabetes mellitus, and smoking) and perioperative cardiovascular complications in a large cohort of patients undergoing non-cardiac surgery in the United States.
Methods:
We performed a retrospective cohort study of patients undergoing non-cardiac surgery between 2009 and 2015, using data from the National Surgical Quality Improvement Program (NSQIP). The NSQIP is a large multicenter surgical registry from the United States that collects data on over 150 perioperative variables from patients undergoing surgery at over 600 centers. At each site, certified nurse reviewers prospectively enroll a systematic sample of the first 40 patients on the operating log over an 8-day cycle. Patients are followed after hospital discharge up to 30 days postoperatively. The accuracy and reproducibility of NSQIP data have been previously validated.9 10 We included data from all adults age ≥ 18 undergoing major non-cardiac surgery in both the inpatient and outpatient setting; patients undergoing cardiac surgery and interventional radiology procedures were excluded (n=24,418). Elective and emergent surgeries were included. Cases with incomplete data (n=6084) or outlier body mass index data (n=58,563) were also excluded, as described below and as shown in Supplemental Figure 1. Eligible non-cardiac surgeries were categorized by the specialty of the operating surgeon into nine subgroups: general surgery, gynecology, neurosurgery, orthopedics, otolaryngology, plastic surgery, thoracic, urology and vascular surgery.
Demographic data, including age, sex, and race/ethnicity, were obtained from the NSQIP. Body mass index (BMI) was calculated from height and weight data. BMIs calculated to be greater than three times the interquartile range above the third quartile or less than the first quartile, which comprised 2.5% of cases, were excluded by this method as outliers that were presumed to be due to inaccurate data entry. Traditional cardiovascular risk factors, including current smoking (defined as use of tobacco cigarettes in the year prior to admission for surgery), hypertension and diabetes mellitus requiring medical therapy, as defined by NSQIP, were determined for all patients undergoing non-cardiac surgery. Smoking, diabetes and hypertension were selected as they are routinely assessed, familiar to patients, and validated as dichotomous risk factors for cardiovascular disease.11 Patients were subdivided into current and former smokers, stratified by total number of pack-years of smoking. Patients with diabetes mellitus were divided into cohorts receiving insulin and those receiving oral medications without insulin. Other clinical covariates of interest included end stage renal disease requiring dialysis, history of congestive heart failure, and history of chronic obstructive pulmonary disease. Pre-operative American Society of Anesthesiologist (ASA) Physical Status classifications were also reported.
The primary outcome of this analysis was perioperative myocardial infarction, occurring intra-operatively or within 30-days post-operatively. Myocardial infarction was defined as any one of the following: documentation of ECG changes indicative of acute MI, new elevation in troponin greater than 3 times upper level of the reference range in the setting of suspected myocardial ischemia, or physician diagnosis of myocardial infarction. ECG changes indicative of acute MI included ST elevation >1 mm in two or more contiguous leads, new left bundle branch block, or new q-waves in ≥2 contiguous leads. Secondary outcomes were all-cause death and the composite outcome of death or MI within 30 days of surgery.
We examined the association between traditional cardiovascular risk factors and the risk of the primary and secondary outcomes in several ways. First, traditional cardiovascular of risk factors were converted into dichotomous variables to identify: (a) current/recent smokers (versus individuals who did not smoke during the one year prior to surgery), (b) patients with diabetes mellitus treated with medication (versus patients not treated with medications for diabetes mellitus), (c) patients with chronic hypertension requiring treatment (versus patients not treated with antihypertensive medications). Univariate associations between risk factors and outcomes were calculated overall, within age and sex subgroups, and by non-cardiac surgical subtype. Associations between risk factors and outcomes were also explored using multivariable logistic regression models adjusted for age as a continuous variable, sex, race, and surgical subtype. To develop a novel risk score, each traditional cardiovascular risk factor was assigned an integer point value based on the log coefficients from multivariable regression analysis. The total risk score for each patient was then calculated by summing the points assigned to each risk factor present.
To establish a potential dose-dependent effect of traditional cardiovascular risk factors, we examined the effect level of smoking exposure, stratified by pack years (<5, 5–10, 10–20, 20–50, 50+ pack years) and by former and current tobacco use, on outcomes using a multivariate-adjusted model. In these analyses, non-smokers were defined as patients who were not current smokers and had zero pack-years of smoking recorded. We also evaluated the effect of diabetes mellitus severity, stratified based on the requirement for oral therapy or insulin for glycemic control. The use of insulin for glycemic control was considered a surrogate for advanced diabetes mellitus.
We calculated population attributable risk (PAR) and 95% confidence intervals (CI) to estimate the proportion of perioperative MI occurring in the population that would not have occurred (assuming a causal relationship) if no traditional cardiovascular risk factors were present. PAR was calculated using the Levin formula, which incorporates the estimates of the adjusted odds ratios (aOR) for each risk factor [PAR = prevalence × (aOR-1) / 1+ (prevalence × (aOR −1))].12 Attributable risk analyses were also performed for death and the composite endpoint of death or MI.
To quantify the association between traditional cardiovascular risk factors and the time from surgery to the first perioperative cardiovascular events, a time to event analysis was conducted as described by Fine and Gray,13 with MI as the outcome event and death as the competing risk. Kaplan-Meier curves were generated to plot the incidence of MI over time. Cox-Proportional hazards models adjusted for age, sex, race, and surgical subtype were used to calculate hazard ratios and 95% confidence intervals for the outcome of MI independent of the risk of death.
A sensitivity analysis was performed to determine associations between traditional risk factors and outcomes after excluding patients undergoing emergency surgery, in whom competing perioperative risks might attenuate associations between traditional cardiovascular risk factors and outcomes. As perioperative risks increase with age, we performed a sensitivity analysis of surgical patients ≥45 years old to confirm associations in a high-risk cohort frequently used in randomized trials of individuals undergoing non-cardiac surgery.14,15 We also performed subgroup analyses by sex, surgery type, and ASA classification to confirm the observed associations despite differences in cardiovascular risk in select populations.
We assessed the efficacy of our simplified risk stratification by comparing it to the Revised Cardiac Risk Index (RCRI), described in the Supplement.3,4,6 RCRI scores were calculated for all NSQIP surgical cases with sufficient data. Due to changes in NSQIP data collection methods, history of ischemic heart disease and stroke were only collected between 2009–2010, limiting RCRI analyses to these years. The incidence and odds of perioperative myocardial infarction (adjusted for age, sex and race) were determined by RCRI score. The negative predictive value for the model was calculated based on the absence of any RCRI criteria. C-statistics were reported to compare the discriminative power of the models as applied to the NSQIP surgical dataset (see Supplemental Fig S3).16
Results:
A total of 3,872,345 non-cardiac surgeries met study inclusion and exclusion criteria between 2009 and 2015. A majority of patients (59%) had ≥1 cardiovascular risk factor at the time of surgery, with 1 risk factor in 40%, 2 risk factors in 17%, and 3 risk factors in 2% (Table 1). Patients with a greater number of cardiovascular risk factors were more frequently male, tended to be older, had elevated BMIs, and higher ASA classifications. A majority of patients underwent general (52%), orthopedic (17.5%), and vascular (7.8%) surgery, with a higher proportion of vascular surgery among patients with multiple risk factors (Table 1).
Table 1:
Characteristics of Patients Undergoing Non-cardiac Surgery According to Number of Pre-operative Traditional Cardiovascular Risk Factors
| 0 RF N =1,594,063 N (%) |
1RF N =1,553,022 N (%) |
2RF N =647,566 N (%) |
3RF N =77,694 N (%) |
|
|---|---|---|---|---|
| Male sex | 596,835 (38) | 686,761 (45) | 317,462 (49) | 41471 (53) |
| Age, mean (SD) | 50 (17) | 60 (16) | 63 (12) | 60 (11) |
| Race | ||||
| White | 117658 (74) | 1180959 (77) | 470601 (73) | 54500 (71) |
| Black | 117652 (7) | 154543 (10) | 93339 (15) | 14502 (19) |
| Asian | 51957 (3) | 35543 (2) | 15692 (2) | 1200 (2) |
| American Indian | 8493 (<1) | 9467 (<1) | 4417 (<1) | 744 (1) |
| Native Hawaiian or Pacific Islander | 6035 (<1) | 5631 (<1) | 3060 (<1) | 367 (<1) |
| Unknown/Not Reported | 216976 (14) | 149231 (10) | 53524 (8) | 5534 (7) |
| Ethnicity | ||||
| Hispanic | 142080 (9) | 88586 (6) | 43718 (7) | 4732 (7) |
| BMI, mean (SD) | 29 (7) | 30 (7) | 32 (8) | 32 (8) |
| History of COPD | 21710 (1) | 79669 (5) | 63833 (10) | 13791 (18) |
| History of CHF | 2401 (<1) | 13061 (1) | 12371 (2) | 2172 (3) |
| Dialysis | 4836 (<1) | 22062 (1) | 25096 (4) | 3450 (4) |
| ASA Class | ||||
| 1 | 310110 (20) | 38042 (2) | 1533 (<1) | 94 (0.01) |
| 2 | 902642 (57) | 702462 (46) | 145898 (23) | 9723 (13) |
| 3 | 701646 (21) | 701646 (46) | 412331 (64) | 53144 (69) |
| 4 | 28466 (2) | 28466 (6) | 79550 (12) | 13602 (18) |
| 5 | 1298 (<1) | 3004 (<1) | 1918 (<1) | 288 (<1) |
| Surgery Type | ||||
| General Surgery | 894268 (56) | 785014 (51) | 303770 (47) | 32315 (42) |
| Gynecology | 144656 (9) | 87062 (6) | 23284 (4) | 1927 (3) |
| Neurosurgery | 61907 (4) | 73242 (5) | 31274 (5) | 3740 (5) |
| Orthopedics | 254286 (16) | 303279 (20) | 105970 (17) | 9443 (12) |
| Otolaryngology | 52056 (3) | 34704 (2) | 11359 (2) | 1251 (2) |
| Plastics | 61079 (4) | 29414 (2) | 7366 (1) | 685 (1) |
| Thoracic | 14238 (1) | 20124 (1) | 9783 (2) | 1424 (2) |
| Urology | 66576 (4) | 86157 (6) | 36679 (6) | 3875 (5) |
| Vascular | 4221 (3) | 122819 (8) | 113926 (18) | 22547 (29) |
| Other Surgery | 15 (<1) | 16 (<1) | 7 (<1) | 1 (<1) |
ASA: American Society of Anesthesiologists
BMI: body mass index
COPD: chronic obstructive pulmonary disease
CHF: congestive heart failure
RF: risk factor
SD: standard deviation
The overall incidence of perioperative MI was 368 per 100,000 non-cardiac surgeries. Death and the composite of death or MI occurred in 1,039 and 1,336 per 100,000 non-cardiac surgeries, respectively. Among patients with perioperative MI, 19% died within the 30-day follow-up period (versus 0.97% mortality at 30-days among patients without MI, p<0.001). Patients undergoing thoracic, vascular, general and neurosurgery had the highest incidence of perioperative death and the composite of death or MI (Figure 1; Supplementary Figure 2).
Figure 1:
Incidence of Perioperative Myocardial Infarction by Presence of Traditional Cardiovascular Risk Factors and Surgery Type
Among all surgeries, the incidence of perioperative MI was higher in patients with diabetes mellitus (826 MIs per 100,000 non-cardiac surgeries), hypertension (651 MIs per 100,000 surgeries), and smoking (444 MIs per 100,000 surgeries) in comparison to those without any traditional CV risk factors (101 MI per 100,000 surgeries, p<0.001 for all comparisons). The secondary endpoints of death or the composite of death or MI followed similar trends (Table 2). The population attributable risk of MI for hypertension, smoking, and diabetes was 65%, 22%, and 5%, respectively (Supplementary Table 1).
Table 2:
Incidence and Adjusted Odds of MI and Perioperative Mortality by Cardiovascular Risk Factor Status
| Perioperative MI (N=14,007) |
Perioperative Mortality (N=38,908) |
Perioperative Death or MI (N=50,264) |
||||
|---|---|---|---|---|---|---|
| N (%) | Adjusted Odds Ratios*
(95% CIs) |
N (%) | Adjusted Odds Ratios*
(95% CIs) |
N (%) | Adjusted Odds Ratios*
(95% CIs) |
|
| Risk Factor | ||||||
| Diabetes N = 590,251 |
4876 (0.83%) |
1.91 (1.85–1.98) |
10806 (1.83%) |
1.41 (1.38–1.45) |
14836 (2.51%) |
1.54 (1.51–1.57) |
| Hypertension N =1,771,206 |
11539 (0.65%) |
2.12 (2.03–2.22) |
28782 (1.62%) |
1.24 (1.21–1.26) |
38128 (2.15%) |
1.39 (1.37–1.42) |
| Smoking N = 719,779 |
3198 (0.44%) |
1.75 (1.68–1.83) |
8325 (1.16%) |
1.77 (1.73–1.82) |
10900 (1.51%) |
1.76 (1.72–1.8) |
| Number of Risk Factors | ||||||
| 0 RF N =1,594,063 |
1616 (0.10%) |
1.0 (Reference) |
7425 (0.47%) |
1.0 (Reference) |
8707 (0.55%) |
1.0 (Reference) |
| 1 RF N =1,553,022 |
6489 (0.42%) |
2.07 (1.96–2.19) |
19273 (1.24%) |
1.34 (1.3–1.38) |
24475 (1.58%) |
1.46 (1.43–1.5) |
| 2 RF N = 647,566 |
5308 (0.82%) |
3.63 (3.43–3.85) |
11974 (1.85%) |
1.89 (1.83–1.94) |
16309 (2.52%) |
2.19 (2.13–2.25) |
| 3 RF N = 77,694 |
836 (1.08%) |
5.54 (5.09–6.04) |
1564 (2.01%) |
2.58 (2.44–2.73) |
23257 (2.90%) |
3.11 (2.96–3.26) |
Adjusted for age, sex, race and surgical subtype
Full coefficients and odds ratios for all adjusted variables in Supplementary table 8
MI: myocardial infarction
RF: risk factors
The incidence of perioperative MI increased in patients based on the number of traditional risk factors present, from 101 MIs per 100,000 surgeries in patients with zero risk factors to 1076 MIs per 100,000 in patients with all three risk factors. The incidence of death and the composite of death or MI also followed similar trends (Table 2). A model weighting each of the three traditional cardiovascular risk factors to generate a perioperative risk score had a C-statistic of 0.695 (95% CI 0.683–0.704) to predict perioperative MI (Supplementary Figure 3). The negative predictive value of perioperative MI in patients with zero risk factors was 99.89% (95% CI 99.89–99.90) in the overall cohort and 99.6% in the subset of patients age ≥45 years.
After adjustment for age, sex, race, and surgery type, traditional cardiovascular risk factors, alone and in combination, significantly increased odds of MI, death and the composite of death or MI. In comparison to patients with zero risk factors, those with 1 risk factor had a two-fold increased odds of 30-day MI (adjusted OR [aOR] 2.07, 95% CI 1.96–2.19), patients with 2 risk factors had 3.6-fold increased odds (aOR 3.63, 95% CI 3.43–3.85), and patients with 3 risk factors had 5.5-fold increased odds (aOR 5.54, 95% CI 5.09–6.04; Table 2). Multiple cardiovascular risk factors also conferred increased odds of all-cause mortality and the composite of death or MI at 30 days (Table 2).
Kaplan-Meier curves depicting cumulative incidence of perioperative myocardial infarction are shown in Figure 2. The adjusted hazard of MI, excluding the competing risk of death, was 2.18 (95% CI 2–2.38) for hypertension, 2.12 (95% CI 1.97–2.27) for diabetes mellitus, and 2.12 (95% CI 1.96–2.31) for smoking. In a competing risk model, the presence of 1 risk factor (adjusted HR [aHR] 2.34, 95% CI 2.01–2.73), 2 risk factors (aHR 4.18, 95% CI 3.57–4.9), and 3 risk factors (aHR 7.15, 95% CI 5.67–9.01) were associated with increasing hazard for perioperative MI.
Figure 2:
Kaplan Meier Cumulative Incidence Curves by Cardiovascular Risk Factors for the Outcome of Perioperative MI with Death as a Competing Risk
KM curves for hypertension (HTN, panel A), diabetes (DM, panel B), current smoking (Smoker, panel C), and cardiovascular risk factor score (RF, panel D). Cumulative number of events are displayed in the table below. Hazard ratios for MI with competing risk of death adjusted for age, sex, race and surgery subtype (aHR) are displayed within the graph.
There was a dose-dependent association between smoking exposure and the risk of perioperative MI in both current and former smokers (Figure 3). Current smokers with more than 10 pack-years exposure and former smokers with more than 20 pack-years exposure had a significantly increased risk of perioperative MI (Supplementary Table 2). In patients with diabetes mellitus, disease severity was also associated with increased risk of perioperative MI. Compared to patients without diabetes mellitus, patients with diabetes mellitus treated with oral medications had a 49% greater risk of perioperative MI (aOR 1.49 95% CI 1.42–1.56) and those who required insulin had a 2.6-fold greater risk of perioperative MI (aOR 2.60 95% CI 2.49–2.72). The association between smoking status and diabetes status with secondary endpoints of death or the composite of death or MI are detailed in Supplementary Table 2.
Figure 3:
Odds of Perioperative MI by Smoking Status and Duration of Exposure*
*Adjusted odds for age, race, sex and surgery type
PY: Pack-years
Subgroup and Sensitivity Analyses:
After excluding patients undergoing emergency surgery, traditional risk factors were associated with perioperative MI, death, and the composite of death or MI. Findings were also consistent in patients age >45 years old and when stratified by sex (Supplementary Table 3). In the subgroup of patients undergoing vascular surgery, traditional risk factors were associated with preoperative MI. Findings were consistent in patients stratified by ASA classification (Supplementary Table 4).
Risk Score Comparisons:
RCRI scores were calculated for 551,667 surgical cases in which all necessary patient characteristics were available (Supplementary Table 5). A majority of these patients (77%) had zero RCRI risk factors. The incidence and adjusted odds of perioperative MI increased in a stepwise fashion with each additional risk factor (Supplementary Table 5). The C-statistic for the RCRI score was 0.701 (95% CI 0.691–0.710) (Supplementary Figure 2).
Discussion:
In this large retrospective cohort study of patients undergoing non-cardiac surgery, three traditional cardiovascular risk factors were strongly and independently associated with risk of perioperative MI, death, and the composite of death or MI. The intensity of diabetes therapy and the extent and timing of tobacco exposure demonstrated a graded relationship with perioperative cardiovascular events. The cumulative number of CV risk factors was associated with large stepwise increases in the odds of perioperative MI (aOR 2.06, 3.63, 5.54 for 1, 2, and 3 risk factors, respectively). Odds of perioperative death increased in a more linear fashion (aOR 1.34, 1.89, 2.58, for 1, 2, and 3 risk factors, respectively). These findings suggest that the presence of diabetes, hypertension and smoking are more strongly associated with perioperative MI than all-cause mortality. Myocardial infarction is a superior marker of modifiable perioperative cardiovascular risk, as a majority of perioperative deaths are ultimately attributed to non-cardiovascular causes.17
In the present study, perioperative MI occurred in 0.36% of surgeries overall and 0.48% of surgeries among adults age ≥45. Although incidence of perioperative cardiac events varies depending on the definition used,7 our findings are similar to other large contemporary databases that use a similar definition and report perioperative MI in ~1% of surgeries.3,4,6,18. Although prospective trials report a higher incidence of post-operative ischemic events, including 3–5% risk of MI in the POISE trial and 7.9% risk of MINS in the VISION trial, these patient populations had a higher burden or cardiovascular disease and underwent rigorous monitoring for myocardial injury. Standardized protocols for troponin measurements were not standard of care during the time of data collection, thus our study is more reflective of a broad, lower risk surgical population undergoing routine post operative care. Notably, subgroup analysis of older patients, patients with comorbid disease and in patients undergoing higher risk vascular surgery, our results were similar (Supplementary tables 3 and 4).
We found that nearly 1 in 5 patients with perioperative MI died within the 30-day follow up period. In a large national readmission database of adults admitted for major non-cardiac surgery, nearly one third of patients identified as having perioperative MI died during the index hospitalization or were readmitted within 30 days of discharge.19 Based on the morbidity and mortality associated with perioperative MI, evaluation of perioperative cardiovascular risk is recommended prior to non-cardiac surgery.7,8 While prior risk scores have attempted to identify patients at the highest risk for cardiovascular events, data from the present analysis suggest that the absence of 3 traditional cardiovascular risk factors may be sufficient to identify low risk groups of patients who are unlikely to have cardiovascular complications of surgery. For patients at low risk, additional cardiac work up appears unnecessary. Despite this, United States Medicare data shows that as many as 3.75% of patients undergo unindicated preoperative cardiac stress testing, and overall rates of preoperative cardiac stress testing has increased by 5% in the past decade.20 Our simplified risk score may reassure providers that in the vast majority of patients without hypertension, diabetes mellitus, or smoking, additional pre-operative risk stratification and cardiovascular testing may not be necessary.
Although multiple perioperative cardiovascular risk assessment scores exist, they have several limitations. The RCRI, Myocardial Infarction and Cardiac Arrest (MICA) and American College of Surgeon’s Surgical Risk Calculator (ACS-SRC) scores all require laboratory data, which may be unavailable in the outpatient setting at the time of pre-operative assessment.3 The MICA and ACS-SRC both rely on ASA classification, which typically involves subjective assessment by an anesthesiologist. The ACS-SRC risk score is particularly complicated and includes 21 components. Additionally, the all three risk scores include high-risk surgery as a predictor of perioperative risk.4,6 By using a score comprised of traditional cardiovascular risk factors, practitioners can easily assess risks based on medical history, without knowledge of laboratory data, surgery type, or a formal anesthesiology assessment. We found similar ROC curves between our simplified risk score and the RCRI (Supplementary Figure 2). Based on the present analysis, assessment of traditional risk factors may be sufficient to accurately determine perioperative risk of MI.
There are several limitations to this manuscript. The NSQIP database was designed for quality improvement rather than survey purposes. We are limited by the NSQIP definition of MI which may be subject to misclassification bias given subjective physician diagnosis. Although the NSQIP definition of MI includes new electrocardiographic Q-waves that may capture silent infarctions, standardized protocols for perioperative troponin measurements were not mandated and monitoring was left at the discretion of the individual treating physicians. Myocardial injury after non-cardiac surgery (MINS), defined as troponin elevated above the upper limit of normal, is associated with higher mortality even in the absence of ischemic symptoms.1,14,21 Thus, the NSQIP definition of MI may miss a subset of patients with silent but prognostic perioperative cardiovascular events. Recent Canadian Cardiovascular Society Guidelines recommend daily surveillance with troponin measurements in the first 48–72 hours post-operatively in patients at risk for cardiovascular events, since identification of patients with MINS may prompt changes in post-operative medical management. In addition to the presumed benefit of aspirin and statin therapy, a recent randomized trial of patients with MINS reported that oral dabigatran reduces long-term adverse cardiovascular events.22,23 Unfortunately, patients included in this analysis of NSQIP data underwent surgery prior to the publication of either the CCS perioperative guidelines or the results of the MANAGE trial. Therefore, the results of these studies did not impact clinical care. Still, this study is reflective of the broad surgical population undergoing routine post-operative care without systematic, protocol-driven measurement of troponin. Furthermore, the NSQIP has rigorously evaluated the reliability of data collection. Inter-rater reliability audits reveal overall disagreement rates of 1.56% and post-operative myocardial infarction had intra-rater disagreement in <1% of NSQIP cases.27 The NSQIP definitions of both hypertension and diabetes are based upon medication use and do not include patients with diet and lifestyle-controlled disease. The NSQIP defines hypertension as a prescription of antihypertensive medication for ≥2 weeks in combination with an established diagnosis of hypertension in the medical record. This definition reduces the likelihood of misclassification when an antihypertensive therapy is prescribed for another cardiac indication, such as beta-blocker or calcium channel blocker use for rate control in a non-hypertensive patient with atrial fibrillation. Still, any misclassification would categorize patients with disease to the control group, biasing comparisons toward the null hypothesis. Furthermore, prior studies have used medication use as a marker for disease prevalence.15 We were unable to assess severity of hypertension as the dataset lacks information on medications or blood pressure measurements. Additionally, current smokers were defined as those who reported smoking within one year of surgery, as per the NSQIP definitions. Despite these limitations, we found prevalent diabetes in 15%, prevalent hypertension in 45% and smoking in 18% of our sample, consistent with other studies (Supplementary Table 6).24 In the current analysis, the Levin formula was used to calculate PAR. This liberal approach may yield higher PAR than other methodologies. Finally, age is an established predictor of post-operative adverse cardiovascular events, and frail patients undergoing non-cardiac surgery require special consideration.23,25 Unfortunately, the NSQIP does not collect data on measures of frailty, and therefore this was not addressed in our analysis. In the current analysis, the absolute risk of perioperative MI increased with age. Still, the negative predictive value associated with the absence of traditional cardiovascular risk factors remained high (99.6%) in patients age ≥45 years.25,26 Thus, despite the limitations of NSQIP data, this manuscript provides important insights into the prognostic value of traditional cardiovascular risk factors from a large multi-center surgical cohort that includes highly reliable, prospective data collection and is representative of real-world clinical care.
Conclusion:
Patients with multiple cardiovascular risk factors, including hypertension, diabetes, or tobacco use, are at increased risk of perioperative MI and mortality. In contrast, perioperative MI was rare in patients without cardiovascular risk factors, occurring in approximately 1 in 1,000 surgeries. Patients without traditional cardiovascular risk factors may be able to forgo unindicated perioperative testing, thereby minimizing waste and expediting surgery. Further evaluation is needed to determine the impact of a simplified risk score in the perioperative setting.
Supplementary Material
Acknowledgements:
Author Contributions: Dr’s Wilcox, Smilowitz and Berger had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Wilcox, Smilowitz, Berger.
Critical revision of the manuscript for important intellectual content: Wilcox, Smilowitz, Beckman, and Berger.
Statistical analysis: Wilcox, Smilowitz, Xia.
Supervision: Berger.
Funding/Support:
JSB was supported, in part, by the National Heart, Lung, and Blood Institute of the National Institutes of Health (R01HL114978) and NRS was supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under award 5T32HL098129-09.
Footnotes
Conflict of Interest Disclosures: Dr. Beckman reports personal fees from Astra Zeneca, personal fees from Bristol Myers Squibb, personal fees from Boehringer Ingelheim, personal fees from Merck, personal fees from Antidote Therapeutics, personal fees from Amgen, personal fees from Sanofi, personal fees from Bayer, personal fees from Novartis, outside the submitted work. All other authors report nothing to disclose.
References:
- 1.Botto F, Alonso-Coello P, Chan MT, et al. Myocardial injury after noncardiac surgery: a large, international, prospective cohort study establishing diagnostic criteria, characteristics, predictors, and 30-day outcomes. Anesthesiology. 2014;120(3):564–578. [DOI] [PubMed] [Google Scholar]
- 2.Smilowitz NR, Gupta N, Guo Y, Beckman JA, Bangalore S, Berger JS. Trends in cardiovascular risk factor and disease prevalence in patients undergoing non-cardiac surgery. Heart. 2018;104(14):1180–1186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Gupta PK, Gupta H, Sundaram A, et al. Development and validation of a risk calculator for prediction of cardiac risk after surgery. Circulation. 2011;124(4):381–387. [DOI] [PubMed] [Google Scholar]
- 4.Lee TH, Marcantonio ER, Mangione CM, et al. Derivation and prospective validation of a simple index for prediction of cardiac risk of major noncardiac surgery. Circulation. 1999;100(10):1043–1049. [DOI] [PubMed] [Google Scholar]
- 5.Davis C, Tait G, Carroll J, Wijeysundera DN, Beattie WS. The Revised Cardiac Risk Index in the new millennium: a single-centre prospective cohort re-evaluation of the original variables in 9,519 consecutive elective surgical patients. Canadian Journal of Anesthesia/Journal canadien d’anesthésie. 2013;60(9):855–863. [DOI] [PubMed] [Google Scholar]
- 6.Bilimoria KY, Liu Y, Paruch JL, et al. Development and evaluation of the universal ACS NSQIP surgical risk calculator: a decision aid and informed consent tool for patients and surgeons. J Am Coll Surg. 2013;217(5):833–842. e833. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Fleisher LA, Fleischmann KE, Auerbach AD, et al. 2014 ACC/AHA guideline on perioperative cardiovascular evaluation and management of patients undergoing noncardiac surgery: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Journal of the American College of Cardiology. 2014;64(22):e77–e137. [DOI] [PubMed] [Google Scholar]
- 8.Members ATF, Kristensen SD, Knuuti J, et al. 2014 ESC/ESA Guidelines on non-cardiac surgery: cardiovascular assessment and management: The Joint Task Force on non-cardiac surgery: cardiovascular assessment and management of the European Society of Cardiology (ESC) and the European Society of Anaesthesiology (ESA). European heart journal. 2014;35(35):2383–2431. [DOI] [PubMed] [Google Scholar]
- 9.Abedi NN, Davenport DL, Xenos E, Sorial E, Minion DJ, Endean ED. Gender and 30-day outcome in patients undergoing endovascular aneurysm repair (EVAR): an analysis using the ACS NSQIP dataset. Journal of vascular surgery. 2009;50(3):486–491, 491.e481–484. [DOI] [PubMed] [Google Scholar]
- 10.Fink AS, Campbell DA Jr., Mentzer RM Jr., et al. The National Surgical Quality Improvement Program in non-veterans administration hospitals: initial demonstration of feasibility. Annals of surgery. 2002;236(3):344–353; discussion 353–344. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Andrus B, Lacaille D. 2013 ACC/AHA guideline on the assessment of cardiovascular risk. J Am Coll Cardiol. 2014;63(25 Part A):2886. [DOI] [PubMed] [Google Scholar]
- 12.Berger JS, Hochman J, Lobach I, Adelman MA, Riles TS, Rockman CB. Modifiable risk factor burden and the prevalence of peripheral artery disease in different vascular territories. J Vasc Surg. 2013;58(3):673–681.e671. [DOI] [PubMed] [Google Scholar]
- 13.Fine JP, Gray RJ. A Proportional Hazards Model for the Subdistribution of a Competing Risk. Journal of the American Statistical Association. 1999;94(446):496–509. [Google Scholar]
- 14.Group PS. Effects of extended-release metoprolol succinate in patients undergoing non-cardiac surgery (POISE trial): a randomised controlled trial. The Lancet. 2008;371(9627):1839–1847. [DOI] [PubMed] [Google Scholar]
- 15.Devereaux PJ, Mrkobrada M, Sessler DI, et al. Aspirin in Patients Undergoing Noncardiac Surgery. N Engl J Med. 2014;370(16):1494–1503. [DOI] [PubMed] [Google Scholar]
- 16.Robin X, Turck N, Hainard A, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics. 2011;12(1):77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Vascular Events In Noncardiac Surgery Patients Cohort Evaluation Study I, Devereaux PJ, Chan MT, et al. Association between postoperative troponin levels and 30-day mortality among patients undergoing noncardiac surgery. Jama. 2012;307(21):2295–2304. [DOI] [PubMed] [Google Scholar]
- 18.Smilowitz NR, Gupta N, Ramakrishna H, Guo Y, Berger JS, Bangalore S. Perioperative Major Adverse Cardiovascular and Cerebrovascular Events Associated With Noncardiac Surgery. JAMA Cardiol. 2017;2(2):181–187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Smilowitz NR, Beckman JA, Sherman SE, Berger JS. Hospital Readmission After Perioperative Acute Myocardial Infarction Associated With Noncardiac Surgery. Circulation. 2018;137(22):2332–2339. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Sheffield KM, Stone PS, Benarroch-Gampel J, et al. Overuse of preoperative cardiac stress testing in medicare patients undergoing elective noncardiac surgery. Annals of surgery. 2013;257(1):73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Roshanov PS, Rochwerg B, Patel A, et al. Withholding versus Continuing Angiotensin-converting Enzyme Inhibitors or Angiotensin II Receptor Blockers before Noncardiac Surgery: An Analysis of the Vascular events In noncardiac Surgery patIents cOhort evaluatioN Prospective Cohort. Anesthesiology: The Journal of the American Society of Anesthesiologists. 2017;126(1):16–27. [DOI] [PubMed] [Google Scholar]
- 22.Devereaux P, Duceppe E, Guyatt G, et al. Dabigatran in patients with myocardial injury after non-cardiac surgery (MANAGE): an international, randomised, placebo-controlled trial. The Lancet. 2018;391(10137):2325–2334. [DOI] [PubMed] [Google Scholar]
- 23.Duceppe E, Parlow J, MacDonald P, et al. Canadian Cardiovascular Society Guidelines on Perioperative Cardiac Risk Assessment and Management for Patients Who Undergo Noncardiac Surgery. Can J Cardiol. 2017;33(1):17–32. [DOI] [PubMed] [Google Scholar]
- 24.Benjamin EJ, Virani SS, Callaway CW, et al. Heart Disease and Stroke Statistics—2018 Update: A Report From the American Heart Association. Circulation. 2018;137(12):e67–e492. [DOI] [PubMed] [Google Scholar]
- 25.Whitney K, Azad N, Power B. Perioperative cardiac risk assessment for the frail older adult. Can J Cardiol. 2018;34(3):343. e311. [DOI] [PubMed] [Google Scholar]
- 26.Parlow JL, Duceppe E, Devereaux PJ. Frailty, the Elderly, and the Guidelines on Perioperative Cardiac Risk Assessment and Management in Noncardiac Surgery. Can J Cardiol. 2018;34(3):343.e313. [DOI] [PubMed] [Google Scholar]
- 27.Shiloach M, Frencher SK Jr, Steeger JE, et al. Toward robust information: data quality and inter-rater reliability in the American College of Surgeons National Surgical Quality Improvement Program. J Am Coll Surg. 2010;210(1):6–16. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.