In this analysis of 1.5 million major noncardiac surgeries performed on patients with T2DM, MACCEs occurred in 1.01% of patients with prior-BS compared with 3.25% of patients with morbid obesity. The rate of perioperative MACCEs did not change significantly over time (0.68% to 1.11%) among patients with prior-BS due to reductions in perioperative death. In contrast, the rate of MACCEs increased significantly from 1.15 to 2.04% among patients with morbid obesity. Overall, prior-BS was associated with a decreased risk of perioperative MACCEs after multivariable adjustment (OR = 0.71, 95% CI 0.62–0.81). To our knowledge, this is the largest study to report on national trends in cardiovascular outcomes among T2DM patients with morbid obesity undergoing major noncardiac surgery. Our findings suggest a benefit of bariatric surgery in T2DM patients to reduce the risk of perioperative cardiovascular events during noncardiac surgery.
T2DM, obesity and noncardiac surgery
Patients with obesity and T2DM comprise a large portion of the population undergoing major noncardiac surgery, and the rate is increasing rapidly [5, 17]. The burden of cardiovascular risk factors (including diabetes, obesity, dyslipidaemia and hypertension) and the prevalence of atherosclerotic cardiovascular disease increased over time after analysing 10.5 million patients hospitalized for noncardiac surgery in the US . In the present study, the rate of morbid obesity increased from 7.42% in 2006 to 17.08% in 2014 among T2DM patients undergoing noncardiac surgery, imposing a heavy burden on individuals and health care systems. Surgical patients with T2DM and obesity have more comorbidities than patients without T2DM and obesity. A prospective study included 7565 surgical inpatients and found that diabetes was associated with increased 6-month mortality, major complications and intensive care unit admission . Obesity, accompanied by not only metabolic disorder but also sleep apnoea and hypoventilation syndrome, increased the rate of postoperative respiratory failure, heart failure, prolonged intubation and intensive care unit transfer [20, 21]. A series of observational studies consistently reported that patients with diabetes and obesity have worse postoperative outcomes, including longer operative times and lengths of stay, increased costs and higher rates of infection and readmission [22,23,24,25]. The American College of Cardiology/American Heart Association also included diabetes as an important factor in perioperative cardiovascular risk stratification and management for the cardiovascular evaluation and management of patients undergoing noncardiac surgery .
Bariatric surgery and cardiovascular benefits
Bariatric surgery is the most effective method to achieve substantial and durable weight loss in people with obesity. The procedure presents a low risk of complications and morbidity, significantly improving quality of life and overall survival, particularly by reducing death due to cardiovascular disease . Although higher perioperative risk and more adverse events were observed in coronary artery disease (CAD) patients undergoing bariatric surgery than in non-CAD operated patients , a recent meta-analysis revealed that bariatric surgery reduced long-term mortality in patients with morbid obesity  and reduced morbidity in obese patients above 43 years old . For patients with T2DM, bariatric surgery not only exhibited superiority over medical treatment in achieving improved glycaemic control but also exhibited cardiovascular benefits . Bariatric surgery was reported to reduce hypertension, hyperlipidaemia and cardiovascular risk in T2DM patients  and reduce epicardial fat mass and ameliorate atrial fibrillation . The early reduction in circulating follistatin after bariatric surgery predicted the improvement in insulin sensitivity observed later in morbidly obese individuals with and without T2DM . Furthermore, the preventive effect of bariatric surgery on mortality was maintained for up to 23 years, while the effects of bariatric surgery on comorbidities and hospital admissions increased over time .
However, the influence of bariatric surgery on cardiovascular events during the perioperative period in patients with T2DM has not been investigated. The present study showed that patients with prior-BS had a significantly lower rate of perioperative MACCEs than those with morbid obesity (1.01% versus 3.25%, P < 0.0001). As previously reported, perioperative MACCEs occurred in 3.0% of the general population  and 3.3% of patients with diabetes , which was consistent with the results of T2DM patients with morbid obesity in the present analysis. On the other hand, the significantly lower rate of perioperative MACCEs in the prior-BS group indicated a benefit of bariatric surgery on cardiovascular events during the perioperative period, which correlated with better control of diabetes, dyslipidaemia and hypertension, even in the presence of poor control of smoking, alcohol abuse and drug abuse (Table 1). Furthermore, unadjusted analysis suggested that prior-BS was associated with a reduced risk of all perioperative outcomes, but the association with AIS disappeared after multivariable adjustment, suggesting that the effect of bariatric surgery on AIS was dependent on general characteristics, comorbidities, cardiovascular risk factors and events, while bariatric surgery had an independent effect on MACCEs, death and AMI.
Differences in MACCEs between general and prior-BS diabetic patients
The trends in MACCEs and individual end points among T2DM patients with prior-BS compared with those among T2DM patients with morbid obesity are encouraging. The adjusted percentage change in MACCEs increased by 6% from 2004 to 2013 in diabetic subjects compared with the percent change in those without , while the rate of MACCEs doubled from 2006 (1.15%) to 2014 (2.04%) in T2DM patients with morbid obesity. In contrast, the rate of perioperative MACCEs was approximately 1.0% and did not change significantly over time for T2DM patients with prior BS. These data indicated that bariatric surgery may prevent the adverse MACCE effects associated with morbid obesity by resulting in weight loss, reducing cardiovascular risk factors (uncontrolled diabetes, hyperlipemia and hypertension) and facilitating a return to the characteristics associated with general diabetic patients or improving these characteristics even further. Notably, concerning each individual end point, the reduction in perioperative death (0.85% in 2006 to 0.33% in 2014) among patients with prior-BS was consistent with that of general diabetic patients (adjusted percentage decreased by 14%), while the rate increased significantly over time (0.56% in 2006 to 0.95% in 2014) among patients with morbid obesity. A recent retrospective analysis included 431,480 subjects and reported that perioperative glucose predicted 30-day mortality after surgery, which was linear for noncardiac surgery but nonlinear for cardiac procedures . Better glycaemic control, especially during the perioperative period, was a vital factor associated with the perioperative mortality of diabetic subjects. In the present analysis, patients with prior-BS had better diabetes control (94.90% vs. 88.04%) along with a lower rate of dyslipidaemia (9.74% vs. 13.09%) and hypertension (72.70% vs. 79.62%) (Table 1), which explained the reduced perioperative mortality in part.
Owing to improved control of cardiovascular risk factors, the trends in AMI and AIS of US patients with diabetes are improving in nonoperative settings , but the trends are inconsistent in operative settings . The unfavourable trends in perioperative cardiovascular outcome may be attributed to platelet activation, catecholamine surges, bleeding and inflammation associated with the surgical procedure . Trends in AMI increased both in the prior-BS and morbid obesity groups of T2DM diabetes in the current study, which contrasted with the adjusted reduction of 7% among general diabetic patients . On the one hand, since there was a lower rate of AMI in the prior-BS group than in the morbid obesity group (0.33% vs. 1.13%), bariatric surgery had a positive influence on perioperative AMI by resulting in weight loss, reducing cardiovascular risk factors, alleviating cardiac workload and altering serum adipokine patterns. On the other hand, the positive influence was still limited, and smoking, alcohol abuse and drug abuse may take part in the increased trend in AMI in T2DM patients with prior-BS (Table 1).
Although the incidence of AIS has declined in the overall US population and among diabetic patients in recent decades , the rate still increased in the overall and diabetic populations during the perioperative period [2, 11]. The trend in perioperative AIS among T2DM patients with prior-BS and morbid obesity was consistent with previous observational data, indicating that bariatric surgery was unable to reverse the increasing trend in perioperative AIS in diabetic patients, although it could reduce the perioperative AIS rate compared with that of morbidly obese subjects. The increasing rate of perioperative AIS, both in the overall population and among patients with diabetes and morbid obesity, may be attributable to altered intraoperative haemodynamic management, use of perioperative β blockers and an increasing prevalence of cardiovascular risk factors among surgical patients [37, 38].
There are several limitations in the present study. First, as we used the NIS database, which is based on administrative coding data, reporting bias or coding errors may exist and be ineluctable. Also, since NIS is an inpatient database focusing only on outcomes occurred during hospitalization and information in the NIS database is de-identified for patient privacy, we could not link discharge or long-term outcomes from other sources. Second, the measures of T2DM control and complications are limited, and data on the course of disease, the level of glucose control, C-peptide level, glycosylated haemoglobin, the treatment regimen and insulin use are lacking, which are important predictors of CVD risk for patients with T2DM . In the present study, we adjusted for uncontrolled T2DM and diabetes with complications as a variate in logistic analysis. Third, to avoid mixing T2DM patients with gestational diabetes patients, patients undergoing obstetric and gynaecological surgeries were not included in the present study, which may have excluded some T2DM patients who were not pregnant. Fourth, data on cardiovascular medication, anticoagulants, and perioperative cardiac biomarkers were not available in the NIS database.