Development and Validation of Predictive Model—HASBLAD Score—For Major Adverse Cardiovascular Events During Perioperative Period of Non-cardiac Surgery: A Single Center Experience in China

Background Major adverse cardiovascular events (MACEs) represent a significant reason of morbidity and mortality in non-cardiac surgery during perioperative period. The prevention of perioperative MACEs has always been one of the hotspots in the research field. Current existing models have not been validated in Chinese population, and have become increasingly unable to adapt to current clinical needs. Objectives To establish and validate several simple bedside tools for predicting MACEs during perioperative period of non-cardiac surgery in Chinese hospitalized patients. Design We used a nested case-control study to establish our prediction models. A nomogram along with a risk score were developed using logistic regression analysis. An internal cohort was used to evaluate the performance of discrimination and calibration of these predictive models including the revised cardiac risk index (RCRI) score recommended by current guidelines. Setting Peking University Third Hospital between January 2010 and December 2020. Patients Two hundred and fifty three patients with MACEs and 1,012 patients without were included in the training set from January 2010 to December 2019 while 38,897 patients were included in the validation set from January 2020 and December 2020, of whom 112 patients had MACEs. Main Outcome Measures The MACEs included the composite outcomes of cardiac death, non-fatal myocardial infarction, non-fatal congestive cardiac failure or hemodynamically significant ventricular arrhythmia, and Takotsubo cardiomyopathy. Results Seven predictors, including H emoglobin, C A RDIAC diseases, Aspartate aminotransferase (A S T), high B lood pressure, L eukocyte count, general A nesthesia, and D iabetes mellitus (HASBLAD), were selected in the final model. The nomogram and HASBLAD score all achieved satisfactory prediction performance in the training set ( C statistic, 0.781 vs. 0.768) and the validation set ( C statistic, 0.865 vs. 0.843). Good calibration was observed for the probability of MACEs in the training set and the validation set. The two predictive models both had excellent discrimination that performed better than RCRI in the validation set ( C statistic, 0.660, P < 0.05 vs. nomogram and HASBLAD score). Conclusion The nomogram and HASBLAD score could be useful bedside tools for predicting perioperative MACEs of non-cardiac surgery in Chinese hospitalized patients.