Association between stress hyperglycemia ratio and postoperative major adverse cardiovascular and cerebrovascular events in noncardiac surgeries: a large perioperative cohort study

Abstract

Background

There has been a concerning rise in the incidence of major adverse cardiovascular and cerebrovascular events (MACCE) following noncardiac surgeries (NCS), significantly impacting surgical outcomes and patient prognosis. Glucose metabolism abnormalities induced by stress response under acute medical conditions may be a risk factor for postoperative MACCE. This study aims to explore the association between stress hyperglycemia ratio (SHR) and postoperative MACCE in patients undergoing general anesthesia for NCS.

Methods

There were 12,899 patients in this perioperative cohort study. The primary outcome was MACCE within 30 days postoperatively, defined as angina, acute myocardial infarction, cardiac arrest, arrhythmia, heart failure, stroke, or in-hospital all-cause mortality. Kaplan-Meier curves visualized the cumulative incidence of MACCE. Cox proportional hazard models were utilized to assess the association between the risk of MACCE and different SHR groups. Restricted cubic spline analyses were conducted to explore potential nonlinear relationships. Additionally, exploratory subgroup analyses and sensitivity analyses were performed.

Results

A total of 592 (4.59%) participants experienced MACCE within 30 days after surgery, and 1,045 (8.10%) within 90 days. After adjusting for confounding factors, compared to the SHR T2 group, the risk of MACCE within 30 days after surgery increased by 1.34 times (95% CI 1.08–1.66) in the T3 group and by 1.35 times (95% CI 1.08–1.68) in the T1 group respectively. In the non-diabetes group, the risk of MACCE within 30 days after surgery increased by 1.60 times (95% CI 1.21–2.12) in the T3 group and by 1.61 times (95% CI 1.21–2.14) in the T1 group respectively, while no statistically significant increase in risk was observed in the diabetes group. Similar results were observed within 90 days after surgery in the non-diabetes group. Additionally, a statistically significant U-shaped nonlinear relationship was observed in the non-diabetes group (30 days: P for nonlinear = 0.010; 90 days: P for nonlinear = 0.008).

Conclusion

In this large perioperative cohort study, we observed that both higher and lower SHR were associated with an increased risk of MACCE within 30 and 90 days after NCS, especially in patients without diabetes. These findings suggest that SHR potentially plays a key role in stratifying cardiovascular and cerebrovascular risk after NCS.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12933-024-02467-w.

Introduction

More than 300 million adults worldwide undergo noncardiac surgery (NCS) annually to improve their quality of life [1, 2], with an average overall complication rate ranging from 7–11% [3]. The mortality rate within 30 days after NCS ranges from 0.5 to 2%, with cardiovascular and cerebrovascular diseases being the leading cause of death [4–8]. One-fifth of high-risk patients undergoing NCS will experience one or more major adverse cardiovascular and cerebrovascular events (MACCE) within one year [9]. The incidence of MACCE gradually becomes a significant factor affecting surgical outcomes and patient prognosis [2].

Research has shown that patients with disturbances in glucose metabolism before surgery are at an elevated risk of experiencing MACCE [8, 10]. The incidence of preoperative hyperglycemia is alarming. In a prospective study, 25% of patients with diabetes had fasting plasma glucose (FPG) levels above the defined threshold prior to the operation [11]. Preoperative hyperglycemia may be a result of poor glycemic control in patients with diabetes or a relative increase in glucose levels due to the neuroendocrine system’s stress response in patients with acute illness, known as stress hyperglycemia [12, 13]. It is defined as a transient physiological response to severe illness. While admission blood glucose (ABG) has been recognized as an indicator of stress-induced hyperglycemia, its ability to accurately distinguish acute glucose elevation is limited due to the influence of chronic glycemic conditions [14, 15]. This is particularly evident in patients with diabetes, as their glucose response to stress events is compromised [16]. Recent research suggests that the pathogenesis of stress-induced hyperglycemia may involve the release of inflammatory mediators and abnormalities in insulin signaling pathways. Furthermore, inflammatory responses have been implicated in the development of postoperative cognitive dysfunction [17]. Chronic hyperglycemia in patients with diabetes is a well-proven risk factor for MACCE [18, 19]. In patients without diabetes, the occurrence of stress-induced hyperglycemia increases their susceptibility to adverse outcomes [20, 21].

Acute perioperative hyperglycemia is more crucial than chronic control for adverse outcomes [22–24]. Therefore, evaluating the true preoperative blood glucose level is of utmost importance. The Stress Hyperglycemia Ratio (SHR) is an innovative marker that provides an estimation of the genuine acute hyperglycemic state by evaluating the acute admission blood glucose levels of the underlying chronic glycemic values [25]. A recent study has confirmed the SHR as an excellent indicator of ‘true stress hyperglycemia’ during hospitalization, suggesting its potential to predict future risks regardless of whether the patient has diabetes [26–30]. Therefore, we hypothesize that there is a correlation between the level of SHR and the occurrence of postoperative MACCE in non-cardiac surgical patients, suggesting that SHR may play a role in risk stratification for MACCE after surgery. Currently, the relationship between SHR and postoperative MACCE has not been studied. Our research objective is to evaluate the association between preoperative SHR and the occurrence of MACCE within 30 and 90 days postoperatively in patients who are undergoing noncardiac surgery, which has not been previously investigated. Furthermore, we sought to determine whether SHR may have any potential clinical significance in this patient population.

Method

Study design and participants

INSPIRE is a large-scale perioperative medical research database. The existing study has provided a comprehensive overview of this database [31]. It recruited a total of 274,919 patients who underwent various surgeries and received anesthesia management at Seoul National University Hospital from January 2011 to December 2020, resulting in a total of 376,839 surgical cases. Detailed records were maintained for each surgical procedure, including surgical and anesthesia-related variables, perioperative diagnoses, vital signs, and laboratory results. After excluding patients under the age of 18 or over 90 on the day of surgery, as well as those who declined to disclose their admission status or had publicly disclosed information, a total of 131,109 surgical records were made publicly available. More detailed information about the database can be found in the PhysioNet platform. An approved researcher (Zhihan, Lyu) was responsible for data extraction. We affirm that all data can be accessed publicly in INSPIRE (version 1.2). This study was approved by Seoul National University Hospital (No. H-2210-078-1368). In consideration of the retrospective nature of the study design, the Institutional Review Board (IRB) has granted a waiver for informed consent. Furthermore, the Institutional Data Review Board at SNUH has reviewed the dataset and deemed it adequately de-identified, leading to their approval for its public release (BRB No. BD-R-2022-11-02).

For patients who underwent multiple surgeries during the study period, we only considered the data from their initial surgery for inclusion in our analysis. Based on this criterion, we initially identified all noncardiac surgeries performed under general anesthesia. Subsequently, we excluded organ transplant surgeries, obstetric surgeries, surgeries with ASA scores of 5 or 6, as well as surgeries where the calculation of SHR was not feasible. Moreover, considering that anemia may affect HbA1c levels, patients diagnosed with anemia before surgery were also excluded. Based on preoperative diagnosis and/or an HbA1c level of 6.5% or higher, the study population was categorized into diabetes and non-diabetes groups.

Data collection

The surgical and anesthesia-related variables, diagnoses, vital signs, and laboratory results were extracted from the clinical data repository. Clinical status variables, such as weight, height, and so on, were measured using physical methods. To minimize the risk of re-identification, each diagnosis was converted to ICD-10-CM. For each procedure, the procedure names were converted to ICD-10-PCS using manual mapping. Patient vital signs were automatically recorded every minute in the anesthesia record. To protect patient privacy as much as possible, the maximum resolution of vital signs was processed to 5 min. Additionally, the anesthesia record included manual entries for urine output, estimated blood loss, volume of fluids or blood products transfused, and values from specialized monitoring devices [31].

Exposure and outcomes

As a new indicator, SHR has been extensively used to assess stress-induced hyperglycemia. We obtained the initial blood glucose and glycated hemoglobin measurements from the participants upon admission. SHR was calculated by the following formula: SHR = ABG (mg/dl) / (28.7 * HbA1c (%) − 46.7) [25]. To ensure a clearer observation of the relationship between SHR and the results, as well as to improve the applicability and simplicity of the model, we divided the SHR into three groups based on the tertiles.

The primary outcome event was the occurrence of MACCE within 30 days after surgery, and the secondary outcome event was the occurrence of MACCE within 90 days after surgery. According to previous studies, we defined MACCE as angina, acute myocardial infarction, cardiac arrest, arrhythmia, heart failure, stroke, in-hospital all-cause mortality, or any combination of them [32, 33]. Each disease was diagnosed by physicians and classified according to the ICD-10-CM.

Variables

The following potential confounding factors were included in the analysis: age (< 45 years/≥45 years), sex (male/female), BMI (kg/m²), Charlson comorbidity index (0–1/≥2), revised cardiac risk index (< 3/≥3), surgery duration (minutes), surgery approach (open/endoscopic), surgery type, emergency surgery, creatinine (mg/dl), anesthesia technique (total intravenous anesthesia/ combined intravenous and inhalation anesthesia), estimated blood loss (ml), urine output (ml), intraoperative crystalloid infusion volume (ml), intraoperative colloid infusion (yes/no), intraoperative blood product transfusion (yes/no), intraoperative vasopressor infusion (yes/no), intraoperative hypotension occurrence (yes/no), and intraoperative hypoxemia occurrence (yes/no). To mitigate the impact of multiple confounders on intraoperative single measurements, intraoperative hypotension was defined as invasive or non-invasive mean arterial pressure below 60mmHg for two consecutive measurements or more [34]. Intraoperative hypoxemia was defined as peripheral oxygen saturation below 90% for two consecutive measurements or more [35]. Among all confounding factors, the continuous variable BMI had a missing rate of 1.32%, which was addressed using the ‘mice’ package for multiple imputation [36].

Statistical analyses

All eligible participants were included in the analysis, and stratified analyses were performed for participants with and without diabetes. Categorical variables were presented as frequencies (percentages) and group differences were assessed using the chi-square test. For continuous variables that passed the normality test, the mean ± standard deviation was used to represent the data, and group differences were evaluated using analysis of ANOVA. For continuous variables that did not pass the normality test, the median (interquartile range, IQR) was used to represent the data, and group differences were assessed using the Kruskal-Wallis rank sum test.

Kaplan-Meier curves were used to visualize the cumulative incidence of outcome events over time in each population. Differences between groups were assessed using the log-rank test. Cox proportional hazard models were then employed to estimate the hazard ratios and 95% confidence intervals for the risk of 30 days and 90 days MACCE, with the median SHR as the reference. Four progressively adjusted Cox proportional hazards models were used. Model 1 was the unadjusted model. Model 2 adjusted for sex, age, and BMI. Model 3 was further adjusted for the Charlson comorbidity index, revised cardiac risk index, surgery approach, surgery type, and anesthesia technique based on Model 2. Model 4 additionally adjusted for surgery duration, emergency surgery, creatinine, estimated blood loss, urine output, intraoperative crystalloid volume administered, intraoperative received colloid infusion, intraoperative received blood product transfusion, intraoperative received vasoactive drug infusion, intraoperative hypotension occurrence, and intraoperative hypoxemia occurrence based on Model 3. Schoenfeld residual tests and visual inspection were used to test the proportional hazards assumption for each model. Multivariable-adjusted restricted cubic splines based on Cox proportional hazards models were used to explore the dose-response relationship between SHR and outcomes in each group. To strike a balance between the flexibility of the fit and the complexity of the model, we strategically placed nodes at the 5th, 35th, 65th, and 95th percentiles. Furthermore, we conducted additional analyses to evaluate the incremental predictive value of SHR over conventional risk factors. Specifically, we calculated the Net Reclassification Index (NRI) and Integrated Discrimination Improvement (IDI) to assess the extent to which SHR improves risk prediction beyond established risk factors.

In exploratory analyses, likelihood ratio tests were performed to compare models with and without cross-product terms, aiming to explore potential interaction effects among subgroups of all participants. In addition, we conducted a series of sensitivity analyses. Firstly, in each analysis, we constructed four models, progressively adjusting for different covariates to demonstrate the robustness of the results. Secondly, we further excluded participants who had received Continuous Renal Replacement Therapy (CRRT) and Extracorporeal Membrane Oxygenation (ECMO) treatment in the six months prior to the study based on the original inclusion criteria. Thirdly, we removed in-hospital all-cause mortality from the original composite outcome definition to conduct another analysis. Fourthly, we performed a reanalysis using logistic regression models to describe the strength of association between SHR and MACCE using odds ratios (ORs). Finally, we calculated the E-value to assess the potential impact of unmeasured confounders on the causal conclusions of the study. A higher E-value indicates a stronger association of unmeasured confounders that would be required to explain the observed effect.

Data processing, analysis, and result visualization were conducted using R software (version 4.3.3). A p-value (two-sided) < 0.05 was considered statistically significant.

Results

Baseline characteristics

In this retrospective study, a comprehensive cohort of 12,899 surgical patients were selected based on rigorous inclusion and exclusion criteria. The study consisted of 6,609 female patients (51%) and 6,290 male patients (49%) in total. A substantial majority of the patients (86%) were aged 45 years or older, indicating a significant age distribution within the cohort. The diabetes status of the study participants revealed that 9,228 patients (72%) were free from diabetes, while 3,671 patients (28%) had a confirmed diagnosis of diabetes. In terms of surgical procedures, the majority of the patients (77%) underwent open surgeries, whereas a minority (23%) chose endoscopic procedures. Table 1 displays the baseline characteristics of the study population, which was stratified into three groups based on tertile levels of the SHR. Supplementary Table S1 and Table S2 present the characteristics of the population classified according to the outcomes.

Table 1.

Baseline characteristics of the study population by tertiles of SHR

Variable	Overall, N = 12,899	T1 (0.36–0.86), N = 4,306	T2 (0.86–1.03), N = 4,314	T3 (1.03–3.13), N = 4,279	P value
Sex, n (%)					< 0.001
Female	6,609 (51%)	2,366 (55%)	2,277 (53%)	1,966 (46%)
Male	6,290 (49%)	1,940 (45%)	2,037 (47%)	2,313 (54%)
Age, n (%)					< 0.001
< 45	1,762 (14%)	490 (11%)	792 (18%)	480 (11%)
≥ 45	11,137 (86%)	3,816 (89%)	3,522 (82%)	3,799 (89%)
Body mass index (kg/m2), Mean ± SD	24.10 ± 3.60	24.16 ± 3.63	24.03 ± 3.58	24.11 ± 3.60	0.254
Medical history
Myocardial infarction, n (%)	323 (2.5%)	116 (2.7%)	88 (2.0%)	119 (2.8%)	0.055
Congestive cardiac failure, n (%)	201 (1.6%)	68 (1.6%)	56 (1.3%)	77 (1.8%)	0.170
Peripheral vascular disease, n (%)	272 (2.1%)	96 (2.2%)	70 (1.6%)	106 (2.5%)	0.018
Cerebrovascular disease, n (%)	683 (5.3%)	245 (5.7%)	157 (3.6%)	281 (6.6%)	< 0.001
Dementia, n (%)	9 (< 0.1%)	4 (< 0.1%)	4 (< 0.1%)	1 (< 0.1%)	0.372
Chronic pulmonary disease, n (%)	308 (2.4%)	96 (2.2%)	89 (2.1%)	123 (2.9%)	0.034
Rheumatological disease, n (%)	35 (0.3%)	8 (0.2%)	10 (0.2%)	17 (0.4%)	0.141
Liver disease, n (%)	550 (4.3%)	168 (3.9%)	167 (3.9%)	215 (5.0%)	0.011
Diabetes mellitus, n (%)	3,671 (28%)	1,278 (30%)	770 (18%)	1,623 (38%)	< 0.001
Hemiplegia or paraplegia, n (%)	15 (0.1%)	4 (< 0.1%)	6 (0.1%)	5 (0.1%)	0.820
Renal disease, n (%)	449 (3.5%)	141 (3.3%)	103 (2.4%)	205 (4.8%)	< 0.001
Any malignancy, n (%)	5,542 (43%)	2,020 (47%)	1,728 (40%)	1,794 (42%)	< 0.001
Metastatic solid tumour, n (%)	293 (2.3%)	96 (2.2%)	81 (1.9%)	116 (2.7%)	0.034
Charlson comorbidity index, n (%)					< 0.001
0 ~ 1	4,621 (36%)	1,394 (32%)	1,900 (44%)	1,327 (31%)
≥ 2	8,278 (64%)	2,912 (68%)	2,414 (56%)	2,952 (69%)
Creatinine (mg/dl), Mean ± SD	0.89 ± 0.57	0.88 ± 0.57	0.86 ± 0.47	0.94 ± 0.64	< 0.001
Blood glucose (mg/dl), Mean ± SD	127.41 ± 48.77	98.51 ± 18.30	112.39 ± 24.70	171.63 ± 56.52	< 0.001
HbA1c (%), Mean ± SD	6.11 ± 1.00	6.27 ± 0.98	5.83 ± 0.86	6.23 ± 1.09	< 0.001
RCRI, n (%)					< 0.001
< 3	12,723 (99%)	4,244 (99%)	4,278 (99%)	4,201 (98%)
≥ 3	176 (1.4%)	62 (1.4%)	36 (0.8%)	78 (1.8%)
Surgery characteristics
Surgery duration (min), Median (IQR)	140 (90, 225)	140 (90, 225)	140 (90, 220)	145 (90, 235)	0.115
Surgery approach, n (%)					0.039
Endoscopic	2,962 (23%)	1,033 (24%)	1,000 (23%)	929 (22%)
Open	9,937 (77%)	3,273 (76%)	3,314 (77%)	3,350 (78%)
Emergency surgery, n (%)	872 (6.8%)	241 (5.6%)	218 (5.1%)	413 (9.7%)	< 0.001
Surgery type, n (%)					< 0.001
Breast and skin	289 (2.2%)	99 (2.3%)	85 (2.0%)	105 (2.5%)
Endocrine	359 (2.8%)	124 (2.9%)	116 (2.7%)	119 (2.8%)
General	3,200 (25%)	1,069 (25%)	1,014 (24%)	1,117 (26%)
Neurosurgery	840 (6.5%)	207 (4.8%)	233 (5.4%)	400 (9.3%)
Ophthalmic and Otolaryngological	995 (7.7%)	325 (7.5%)	312 (7.2%)	358 (8.4%)
Orthopedic	711 (5.5%)	179 (4.2%)	233 (5.4%)	299 (7.0%)
Regional	828 (6.4%)	344 (8.0%)	293 (6.8%)	191 (4.5%)
Reproductive	1,607 (12%)	699 (16%)	625 (14%)	283 (6.6%)
Respiratory	504 (3.9%)	141 (3.3%)	127 (2.9%)	236 (5.5%)
Urological	981 (7.6%)	331 (7.7%)	374 (8.7%)	276 (6.5%)
Other	2,585 (20%)	788 (18%)	902 (21%)	895 (21%)
Anesthesia technique, n (%)					0.157
CIIA	6,331 (49%)	2,062 (48%)	2,141 (50%)	2,128 (50%)
TIVA	6,568 (51%)	2,244 (52%)	2,173 (50%)	2,151 (50%)
Estimated blood loss (ml), Median (IQR)	150 (40, 380)	160 (40, 400)	150 (40, 350)	140 (40, 360)	< 0.001
Urine output (ml), Median (IQR)	120 (0, 315)	130 (0, 320)	105 (0, 290)	120 (0, 320)	0.033
Crystalloid volume administered (ml), Median (IQR)	1,000 (500, 1,600)	1,000 (550, 1,685)	1,000 (550, 1,600)	900 (500, 1,600)	< 0.001
Received colloid infusion, n (%)	2,584 (20%)	895 (21%)	889 (21%)	800 (19%)	0.028
Received blood products infusion, n (%)	436 (3.4%)	142 (3.3%)	126 (2.9%)	168 (3.9%)	0.034
Received vasoactive drug infusion, n (%)	6,965 (54%)	2,184 (51%)	2,275 (53%)	2,506 (59%)	< 0.001
Intraoperative hypotension, n (%)	4,671 (36%)	1,596 (37%)	1,523 (35%)	1,552 (36%)	0.234
Intraoperative hypoxemia, n (%)	2,162 (17%)	720 (17%)	700 (16%)	742 (17%)	0.383

IQR: Interquartile Range, RCRI: Revised Cardiac Risk Index, CIIA: Combined Intravenous and Inhalation Anesthesia, TIVA: Total Intravenous Anesthesia

Clinical outcomes for MACCE

Figure 1 illustrates the 30-day and 90-day cumulative incidence rates of MACCE in the entire study population and subgroups based on diabetes status. Both the overall population and the non-diabetic subgroup exhibit a consistent pattern of cumulative incidence rates, with the highest rate observed in the T3 group, followed by the T1 group, and the lowest rate in the T2 group. These differences in cumulative incidence rates are statistically significant (log-rank p < 0.0001). However, there were no statistically significant differences observed in the cumulative incidence rates among patients with diabetes.

Fig. 1.

Kaplan–Meier curves for the cumulative incidence of MACCE after noncardiac surgery. (a) Cumulative incidence of MACCE within 30 days after noncardiac surgery in all participants. (b) Cumulative incidence of MACCE within 30 days after noncardiac surgery in the non-diabetic group. (c) Cumulative incidence of MACCE within 30 days after noncardiac surgery in the diabetic group. (d) Cumulative incidence of MACCE within 90 days after noncardiac surgery in all participants. (e) Cumulative incidence of MACCE within 90 days after noncardiac surgery in the non-diabetic group. (f) Cumulative incidence of MACCE within 90 days after noncardiac surgery in the diabetic group. MACCE: major adverse cardiovascular and cerebrovascular events, SHR: stress hyperglycemia ratio.

During the observation period, a total of 592 patients (4.59%, including 45 deaths) experienced adverse events within 30 days after surgery. Similarly, within 90 days after surgery, 1,045patients (8.10%, including 98 deaths) experienced adverse events. We adjusted the model based on the three groups of SHR, with the T2 group as the reference (Table 2). Initially, we observed a significant correlation between preoperative SHR values and MACCE within 30 days postoperatively in all models among the unstratified surgical patients. However, this relationship lost statistical significance in the fully adjusted models within 90 days. In the unadjusted original analysis model, both the T1 group and T3 group had a higher risk of experiencing MACCE within 90 days after surgery compared to the T2 group [T1: 1.33 (95% CI: 1.13–1.56); T3: 1.66 (95% CI: 1.42–1.93)]. The risk within 30 days after surgery was even higher [T1: 1.54 (95% CI: 1.24–1.92); T3: 1.97 (95% CI: 1.60–2.43)]. In the adjusted model 4, the results for the 30-day postoperative periods remained consistent. Specifically, the T1 group had a hazard ratio of 1.35 (95% CI: 1.08–1.68) within 30 days, while the T3 group had a hazard ratio of 1.34 (95% CI: 1.08–1.66) within 30 days.

Table 2.

The association of SHR with MACCE after noncardiac surgery

Group	Total N	No. of events (Incident rate)	Model 1		Model 2		Model 3		Model 4
			HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value
30 days MACCE
All	12899	592 (4.59)
SHR T1	4306	203 (4.71)	1.54 (1.24, 1.92)	< 0.001	1.46 (1.18, 1.82)	< 0.001	1.34 (1.08, 1.68)	0.008	1.35 (1.08, 1.68)	0.008
SHR T2	4314	133 (3.08)	Ref		Ref		Ref		Ref
SHR T3	4279	256 (5.98)	1.97 (1.60, 2.43)	< 0.001	1.80 (1.46, 2.22)	< 0.001	1.46 (1.18, 1.81)	< 0.001	1.34 (1.08, 1.66)	0.007
Without diabetes	9228	328 (3.55)
SHR T1	3028	114 (3.76)	1.64 (1.23, 2.18)	< 0.001	1.58 (1.19, 2.10)	0.002	1.64 (1.23, 2.18)	< 0.001	1.61 (1.21, 2.14)	0.001
SHR T2	3544	82 (2.31)	Ref		Ref		Ref		Ref
SHR T3	2656	132 (4.97)	2.18 (1.65, 2.87)	< 0.001	2.03 (1.54, 2.67)	< 0.001	1.81 (1.37, 2.39)	< 0.001	1.60 (1.21, 2.12)	0.001
With diabetes	3671	264 (7.19)
SHR T1	1278	89 (6.96)	1.06 (0.75, 1.49)	0.759	1.04 (0.74, 1.47)	0.805	0.99 (0.70, 1.40)	0.952	1.02 (0.72, 1.45)	0.890
SHR T2	770	51 (6.62)	Ref		Ref		Ref		Ref
SHR T3	1623	124 (7.64)	1.16 (0.84, 1.60)	0.377	1.13 (0.81, 1.57)	0.465	1.07 (0.77, 1.49)	0.681	1.02 (0.73, 1.43)	0.896
90 days MACCE
All	12899	1045 (8.10)
SHR T1	4306	350 (8.13)	1.33 (1.13, 1.56)	< 0.001	1.27 (1.08, 1.48)	0.004	1.15 (0.98, 1.35)	0.089	1.14 (0.97, 1.34)	0.114
SHR T2	4314	267 (6.19)	Ref		Ref		Ref		Ref
SHR T3	4279	428 (10.00)	1.66 (1.42, 1.93)	< 0.001	1.52 (1.30, 1.77)	< 0.001	1.21 (1.04, 1.42)	0.015	1.12 (0.96, 1.31)	0.148
Without diabetes	9228	588 (6.37)
SHR T1	3028	196 (6.47)	1.36 (1.10, 1.66)	0.004	1.31 (1.07, 1.61)	0.010	1.36 (1.10, 1.67)	0.004	1.32 (1.07, 1.62)	0.009
SHR T2	3544	171 (4.83)	Ref		Ref		Ref		Ref
SHR T3	2656	221 (8.32)	1.76 (1.44, 2.15)	< 0.001	1.65 (1.35, 2.02)	< 0.001	1.44 (1.17, 1.76)	< 0.001	1.27 (1.03, 1.56)	0.023
With diabetes	3671	457 (12.45)
SHR T1	1278	154 (12.05)	0.97 (0.75, 1.25)	0.795	0.95 (0.74, 1.23)	0.707	0.89 (0.69, 1.15)	0.387	0.90 (0.70, 1.17)	0.427
SHR T2	770	96 (12.47)	Ref		Ref		Ref		Ref
SHR T3	1623	207 (12.75)	1.03 (0.81, 1.31)	0.814	1.00 (0.78, 1.27)	0.968	0.95 (0.74, 1.21)	0.671	0.91 (0.71, 1.17)	0.476

Model 1: unadjusted. Model 2: adjusted for sex, age and BMI. Model 3: model 2 + further adjusted for the Charlson comorbidity index, revised cardiac risk index, surgery approach, surgery type, and anesthesia technique. Model 4: model 3 + further adjusted for surgery duration, emergency surgery, creatinine, estimated blood loss, urine output, intraoperative crystalloid volume administered, intraoperative received colloid infusion, intraoperative received blood product transfusion, intraoperative received vasoactive drug infusion, intraoperative hypotension occurrence, and intraoperative hypoxemia occurrence. MACCE: major adverse cardiovascular and cerebrovascular events, SHR: stress hyperglycemia ratio, HR: Hazard Ratio, CI confidence interval

In addition, when we grouped the population according to diabetes status, the non-diabetic group had 328 individuals (3.55%) experiencing outcome events at 30 days and 588 individuals (6.37%) at 90 days. The diabetic group had 264 individuals (7.19%) experiencing outcome events at 30 days and 457 individuals (12.45%) at 90 days. Interestingly, multivariate Cox analysis with adjustment showed a significant association between SHR and postoperative MACCE in populations without diabetes, but not in the population with diabetes. In the population without diabetes, the T1 and T3 groups exhibited an elevated risk of MACCE within both 30 days [T1: 1.61 (95% CI: 1.21–2.14); T3: 1.60 (95% CI: 1.21–2.12)] and 90 days [T1: 1.32 (95% CI: 1.07–1.62); T3: 1.27 (95% CI: 1.03–1.56)] compared to the T2 group in model 4.

The RCS curves in Fig. 2 demonstrate the dose-response relationship between SHR and MACCE, showing no significant nonlinear relationship in the overall population (30 days: P for nonlinear = 0.322; 90 days: P for nonlinear = 0.565). However, in the population without diabetes, a statistically significant nonlinear U-shaped relationship was observed (30 days: P for nonlinear = 0.010; 90 days: P for nonlinear = 0.008). The risk of MACCE decreased initially and then increased with increasing SHR. No significant dose-response association was observed in those suffering from diabetes. In addition, the performance of the model predicting MACCE using SHR is shown in Table S3. The inclusion of SHR in the traditional model indeed significantly improved the risk reclassification ability.

Fig. 2.

Relationships between SHR and the risk of MACCE according to the restricted cubic spline analysis. (a) RCS curve for the risk of MACCE within 30 days after noncardiac surgery in all participants. (b) RCS curve for the risk of MACCE within 30 days after noncardiac surgery in the non-diabetic group. (c) RCS curve for the risk of MACCE within 30 days after noncardiac surgery in the diabetic group. (d) RCS curve for the risk of MACCE within 90 days after noncardiac surgery in all participants. (e) RCS curve for the risk of MACCE within 90 days after noncardiac surgery in the non-diabetic group. (f) RCS curve for the risk of MACCE within 90 days after noncardiac surgery in the diabetic group. MACCE: major adverse cardiovascular and cerebrovascular events, SHR: stress hyperglycemia ratio, RCS: restricted cubic spline. All analyses were adjusted for confounding factors including sex, age, BMI, Charlson comorbidity index, revised cardiac risk index, surgery approach, surgery type, anesthesia technique, surgery duration, emergency surgery, creatinine, estimated blood loss, urine output, intraoperative crystalloid volume administered, intraoperative received colloid infusion, intraoperative received blood product transfusion, intraoperative received vasoactive drug infusion, intraoperative hypotension occurrence, and intraoperative hypoxemia occurrence.

Subgroup and sensitivity analyses

The results of the subgroup analysis are presented in Fig. 3. The results of subgroup analyses showed that the interaction effects were not statistically significant, suggesting that the findings are robust (all P for interaction > 0.05). Furthermore, sensitivity analyses confirmed the stability of our results (Figure S1 and Table S4-6)

Fig. 3.

Subgroup and interaction analyses between the SHR and MACCE after noncardiac surgery. (a) Subgroup and interaction analyses between the SHR and MACCE within 30 days after noncardiac surgery. (b) Subgroup and interaction analyses between the SHR and MACCE within 90 days after noncardiac surgery. MACCE: major adverse cardiovascular and cerebrovascular events, SHR: stress hyperglycemia ratio. All analyses were adjusted for confounding factors including sex, age, BMI, Charlson comorbidity index, revised cardiac risk index, surgery approach, surgery type, anesthesia technique, surgery duration, emergency surgery, creatinine, estimated blood loss, urine output, intraoperative crystalloid volume administered, intraoperative received colloid infusion, intraoperative received blood product transfusion, intraoperative received vasoactive drug infusion, intraoperative hypotension occurrence, and intraoperative hypoxemia occurrence.

Discussion

In this cohort study, we investigated the association between SHR and postoperative MACCE in patients who have undergone NCS. Our study findings highlight that SHR, as a risk factor, has an impact on the occurrence of MACCE at both 30 days and 90 days postoperatively. This association remained significant even after adjusting for other risk factors such as age, sex, BMI, RCRI, and so on. In the study, we observed that patients in the lowest and highest tertiles of SHR had a 35% and 34% higher risk of experiencing MACCE 30 days after the surgery. Notably, the highest SHR in the nondiabetic group was associated with a 60% increased risk of MACCE within 30 days, whereas the lowest SHR was associated with a 61% elevated in the risk of MACCE. Thus, SHR in patients without diabetes is significantly associated with postoperative MACCE risk. Furthermore, we observed a U-shaped curve relationship within the group without diabetes. We conducted various subgroup and sensitivity analyses, and the results remained consistent. According to our information, this is the first study comparing the impact of different SHR levels on postoperative MACCE occurrence in NCS. The findings emphasize the importance of careful glycemic control in NCS to improve postoperative outcomes, particularly in individuals without diabetes. These people should also proactively control their blood glucose levels. This can be achieved through maintaining a healthy lifestyle, including dietary control, exercise, as well as following medical advice regarding sugar and carbohydrate intake. For non-diabetic patients with excessively high preoperative random blood glucose levels or undergoing high-risk surgeries, screening for HbA1c should be conducted to further calculate the SHR to assess the postoperative risk. During the surgical procedure, continuous glucose monitoring systems (CGM) or regular blood glucose measurements can be utilized, and postoperatively, blood glucose levels should be maintained within a safe range.

Although the association between diabetes and MACCE in patients undergoing NCS is well-established [37], limited attention has been given to the impact of preoperative blood glucose levels. Some guidelines recommend preoperative blood glucose control at the target level for patients with diabetes, but 12-30% of perioperative hyperglycemia patients have no history of diabetes [8, 38]. Additionally, there is widespread controversy regarding the ideal treatment drugs and target blood glucose levels. Surgical patients are often in a state of ‘stress hyperglycemia’ [38]. Prior research has characterized stress hyperglycemia as an elevated ABG level and has identified it as an independent risk factor for unfavorable outcomes in individuals diagnosed with Takotsubo syndrome [39], stroke [40, 41], and acute coronary syndrome [42–44]. Considering the potential influence of patients’ previous blood glucose status, Robert et al. proposed the SHR index [25], which is designed to be independent of background blood glucose levels. Xu et al. discovered a meaningful link between the SHR and the occurrence of mortality during hospitalization among patients diagnosed with coronary artery disease [45]. This finding underscores the potential significance of the SHR index as a predictor of patient outcomes in the hospital setting. It is worth noting that our study results indicate a closer relationship between SHR and postoperative MACCE in patients without diabetes compared to patients with diabetes. Similar differences have also been found in non-surgical studies and surgical studies. In a study utilizing two cohorts from the United States and China, researchers observed that critically ill patients diagnosed with acute myocardial infarction who have a high SHR index face an elevated likelihood of experiencing all-cause mortality [46]. Notably, this association was observed exclusively in non-diabetic groups. Additionally, Yang et al. identified a strong association between SHR and MACCE occurring within 30 days after undergoing percutaneous coronary intervention [47]. In patients with acute coronary syndrome undergoing drug-eluting stent implantation, there was a strong U-shaped correlation between SHR and both MACCE and major adverse cardiovascular events [26]. However, it is worth noting that here, although patients in the Q1245 group had a higher risk of experiencing outcome events, their average HbA1c levels were higher than those in the Q3 group, suggesting a potentially higher proportion of undiagnosed diabetes in the Q1245 group. They did not exclude patients with diabetes. Among individuals who do not have diabetes, acute elevation or fluctuation of blood glucose levels within the normal range may result in heightened endothelial dysfunction and oxidative stress, surpassing the impact of chronic hyperglycemia typically observed in patients who have been diagnosed with diabetes [48]. For subjects without diabetes, elevated glucose concentrations could potentially indicate undiagnosed, unmonitored, and untreated diabetes [49].

Stress-induced hyperglycemia is triggered by the hypothalamic-pituitary-adrenal axis, further exacerbating the state of hyperglycemia [50]. The mechanism is as follows. Many disruptions in the insulin receptor-mediated signaling pathway can lead to the occurrence of insulin resistance [20]. Additionally, there is a significant release of hyperglycemic hormones such as glucagon, cortisol, and catecholamines [51]. Furthermore, the body releases a substantial amount of pro-inflammatory cytokines that damage pancreatic β-cells, induce β-cell apoptosis, and reduce peripheral glucose utilization. It is a transient physiological response to illness. Surgery is considered a stress stimulus for patients. Perioperative hyperglycemia frequently occurs and is attributed to factors such as preoperative medication management, fasting guidelines, and surgical stress response [22]. Among patients undergoing tumor resection, the occurrence of preoperative hyperglycemia is not only attributed to the mechanisms mentioned above but also potentially related to the tumor itself or its treatment. Metabolic disturbances are observed in certain cancers and cachexia patients, leading to alterations in glucose metabolism through the Warburg effect or by promoting hormone secretion. Additionally, tumor-induced chronic systemic inflammation plays a crucial role in cancer-associated cachexia, with pro-inflammatory cytokines such as TNF-α, IL-6, IL-2, IL-8, and IFN-γ playing a vital role and contributing to elevated blood glucose levels. Changes in cellular metabolism mediated by oncogenes can also stimulate inflammatory responses in immune cells within the tumor microenvironment. Surgical tumor resection, on the other hand, may improve the state of hyperglycemia [52]. Compared to absolute fasting blood glucose, a sudden increase in blood glucose exceeding the patient’s normal background levels is more closely associated with prognosis and is likely indicative of stress-induced hyperglycemia [13, 53]. Stress hyperglycemia, representing relative hyperglycemia, has been shown to predict worsened outcomes in patients with severe acute illnesses [25]. It has the potential to induce oxidative stress and inflammatory reactions, resulting in cardiac injury, augmented infarct size, and a heightened risk of mortality [54, 55]. Additionally, impaired nitric oxide bioavailability, increased thrombotic activity, enhanced coagulation system, and reduced fibrinolytic activity are observed in patients, promoting a pro-thrombotic environment and contributing to mortality [56, 57]. Previous studies have indicated a link between elevated SHR and increased thrombus burden, as well as reduced flow grade observed during angiography [58, 59].

Moreover, in some studies, SHR has demonstrated a U-shaped relationship with prognosis, although the underlying mechanisms remain unclear [26]. Studies have indicated that the body’s natural response to stress-induced hyperglycemia, characterized by adaptive glucose elevation, may have beneficial effects [60, 61]. Mild to moderate stress-induced hyperglycemia can increase cardiac output. This adaptive glucose elevation helps optimize the utilization of glucose by cells, reduces apoptosis in myocardial cells, and enables the body to meet the increased metabolic demands that accompany critical illness [62]. This adaptive response is believed to confer a survival advantage when faced with acute stress and critical illness, enabling the organism to more effectively meet the high metabolic demands imposed by these challenging conditions. Therefore, calculating preoperative SHR can effectively identify patients at higher risk for postoperative complications, highlighting the importance of closer monitoring and optimal management for their postoperative recovery. In patients without diabetes of this study, when SHR < 0.93, as SHR gradually increases, mild to moderate stress-induced hyperglycemia may have a protective effect against adverse cardiovascular events. When SHR > 0.93, the gradual increase in SHR may be a true indication of stress-induced hyperglycemia. This suggests that mild to moderate SHR may have a protective effect in patients without diabetes. The reasons why these patients exhibit this phenomenon could be as follows: Firstly, some individuals with previously undiagnosed hyperglycemia who have not been diagnosed or treated for diabetes, may indeed be true diabetes patients. This subgroup represents a high-risk population that has not received diagnosis and treatment. However, we have categorized these patients as the diabetes group as accurately as possible based on their HbA1c levels upon admission. Secondly, it is possible that some hyperglycemic patients without a history of diabetes have underlying insulin resistance [63, 64]. Insulin resistance is considered to have potential mechanisms in the development of cardiovascular and all-cause mortality. This is believed to occur through the direct atherogenic effects of insulin on the vessel wall and/or indirect effects through obesity, blood pressure, blood lipids, and metabolic homeostasis [65]. Finally, different thresholds may apply to different populations or settings. It is also possible that non-diabetic individuals require a greater degree of stress to reach the same hyperglycemic state as diabetic patients.

Patients with preoperative abnormalities in glucose metabolism are at an elevated risk of developing cardiovascular complications. As a result, surgeons, cardiologists, and anesthesiologists can obtain preoperative measurements of admission blood glucose and glycated hemoglobin levels to facilitate the prompt determination of the SHR. This calculation enables an estimation of the actual acute hyperglycemic state, thereby enabling the assessment of the risk of postoperative MACCE, particularly among those without diabetes. Importantly, this assessment remains valid even after accounting for other established preoperative factors. These findings contribute to more comprehensive preoperative discussions concerning prognosis and provide valuable insights for related decision-making processes.

This study examined a substantial cohort of surgical patients and presents novel findings on the dose-response relationship between SHR in surgical patients and the occurrence of MACCE following surgery. Nevertheless, it is crucial to recognize the constraints and shortcomings of this study. Firstly, it is important to note that the study cohort is derived from a single institution and comprised solely of Asian patients. Therefore, caution should be exercised when interpreting these results in a broader context. Secondly, as an observational study focusing on NCS patients, it is important to acknowledge the wide range of surgical procedures and patient characteristics involved. Despite meticulous adjustments for variables such as surgical type, approach, and anesthesia method, it should be noted that there may still be unaccounted confounding factors. Nonetheless, the analysis of the E-value reveals that, under the control of measured confounding factors, the unmeasured confounding effect would need to be at least 2.60 to completely invalidate the Hazard Ratio observed in this study [66]. Thirdly, another limitation of our study is the inability to incorporate the year of surgery as a confounding factor in the analysis. Due to the need to protect patient privacy, all times were converted to times relative to the time zero. We were unable to adjust for the potential influence of advances in medical practice and environmental conditions on the incidence of MACCE. Fourthly, the baseline differences in cerebrovascular disease among groups may weaken the conclusions of the study. This reminds us to interpret the results with caution. Finally, it is worth noting that fluctuations in random glucose concentrations can be influenced by various factors, such as recent dietary intake or timing, as well as inherent variations in glucose metabolism, potentially leading to an underestimation of the observed effects. Further validation through large-scale randomized controlled trials is needed in the future.

Conclusion

In conclusion, our retrospective study reveals a significant independent relationship between SHR and the increased risk of MACCE within 30 and 90 days following NCS. Notably, this association exhibits a distinct U-shaped pattern among individuals who do not have diabetes. These findings underscore that SHR may play a potential role in stratifying postoperative cardiovascular and cerebrovascular risks in the population undergoing noncardiac surgery.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Concept: ZHL. Data curation: ZHL and YHJ. Methodology: YXJ and ZHL. Project administration and resources: ZHL. Manuscript Writing: ZHL and YXJ. Review and editing: ZHL, YHJ and YXJ. Supervision: ZHL. All authors read and approved the final manuscript.

Funding

Not applicable.

Data availability

No datasets were generated or analysed during the current study.

Declarataions

Ethics approval and consent to participate

This study was approved by Seoul National University Hospital (No. H-2210-078-1368).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.