The study population consisted of 1,167,414 non-cardiac surgeries registered in the ACS NSQIP database between 2008 and 2012 (Dakik et al., 2019; Dakik et al., 2020a; Dakik et al., 2020b; User guide for the 2016 ACS NSQIP, 2019). The population was stratified by age (≥ 40 and < 40 years old) and by gender (men and women). The performance of the AUB-HAS2 Index was compared between the two age subgroups and the two gender subgroups. There were 2644 surgeries in which gender classification was missing. Those were included in the age but not in the gender analysis. Datasets after 2012 were not included in our study because they did not capture cardiac history on patients which is one of the essential elements in the AUB-HAS2 Cardiovascular Risk Index. The ACS NSQIP is a large multicenter database that collects data on patients undergoing major surgical procedures from more than 250 participating sites on more than 150 variables, including pre-operative risk factors, intraoperative variables, and 30-day post-operative mortality and morbidity outcomes (Dakik et al., 2020b). The data is collected by trained surgical clinical reviewers at each site using a systematic sampling process and is subject to regular inter-rater reliability audits to assess its quality. Required data variables are entered via web-based data collection to the ACS NSQIP website. Surgeries are entered in the database using the International Classification of Diseases (ICD) codes. Patients under the age of 18 were excluded from the database as well as minor and transplant surgeries.
Each patient was assigned an AUB-HAS2 score of 0, 1, 2, 3, and > 3 based on the number of data elements s/he has. The AUB-HAS2 elements are history of heart disease, symptoms of heart disease (angina or dyspnea), age ≥ 75 years, anemia (hemoglobin < 12 mg/dl), vascular surgery, and emergency surgery. Patients were designated as having a history of heart disease if they had a history of prior myocardial infarction, coronary angioplasty, cardiac surgery, heart failure, atrial fibrillation, or moderate/severe valvular disease confirmed by echocardiography. The primary outcome measure was all-cause mortality, myocardial infarction, or stroke at 30 days after surgery (Dakik et al., 2019; Dakik et al., 2020a; Dakik et al., 2020b). Myocardial infarction was defined by ECG changes indicative of an acute MI (one or more of the following three: ST elevation > 1 mm in two or more contiguous leads, new left bundle branch, or new q-waves in two of more contiguous leads) or new elevation in troponin greater than three times the upper level of the reference range in the setting of suspected myocardial ischemia. Stroke was defined as the new occurrence of a motor, sensory, or cognitive dysfunction which persists for more than 24 h.
Statistical analysis
Descriptive analysis was performed and presented in the respective tables. Categorical variables are presented as number and percentages and continuous variables as mean ± standard deviation. Comparisons of the baseline clinical characteristics between the two age subgroups and between the two gender subgroups were performed using Pearson’s chi-squared test for categorical and the ANOVA test for continuous ones. The performance of the AUB-HAS2 Index within each age and gender subgroup was assessed by comparing the event rates among the different score groups (0, 1, 2, 3, and > 3). The Cochran-Armitage test for trend was used to evaluate the trend in the proportions of the outcome across the levels of the AUB-HAS2 score. Time to event analyses were carried out using the Kaplan-Meier curve, where significance was calculated using the log-rank test. Furthermore, receiver operating characteristic (ROC) curves were constructed, and the areas under the curve (AUC) were measured to assess the discriminatory power of the AUB-HAS2 Index in each subgroup. Comparison of AUCs of the AUB-HAS2 in the two age and gender subgroups was performed using the non-parametric Z test (DeLong et al., 1988). The Statistical Analysis Software (SAS, version 9.4) was used for data management and analyses. Statistical significance was set at the 0.05 alpha level.