Abstract
Background
Stress hyperglycemia is prevalent in acute ST-elevation myocardial infarction (STEMI) patients and associates with adverse outcomes. The stress hyperglycemia ratio (SHR), which standardizes admission glucose against estimated average glucose from HbA1c, reduces chronic glycemic interference. However, SHR's predictive value for heart failure (HF) during hospitalization following emergency percutaneous coronary intervention (PCI) remains unclear.
Methods
This single-center retrospective cohort study enrolled 1,068 consecutive STEMI patients undergoing emergency PCI from January 2020 to June 2025. SHR was calculated as the ratio of admission plasma glucose to estimated average glucose. The primary endpoint was hospitalization HF. Predictive efficacy was evaluated using multivariate logistic regression, propensity score matching (PSM), receiver operating characteristic (ROC) analysis, net reclassification improvement (NRI), and integrated discrimination improvement (IDI).
Results
Overall, 158 patients (14.8%) developed hospitalization HF. The HF group showed significantly higher SHR than non-HF patients [1.52 (1.25–1.88) vs. 1.10 (0.92–1.30); P < 0.001]. Each 1-unit SHR increase associated with 2.31-fold HF risk elevation (adjusted OR=2.31, 95% CI 1.68–3.18; P < 0.001). Following 1:1 PSM, high SHR patients maintained significantly higher HF incidence (20.4% vs. 10.2%; P < 0.001). SHR's ROC area under curve was 0.748, superior to admission glucose (0.728; P = 0.031). Adding SHR yielded NRI = 0.318 and IDI = 0.035 (both P < 0.001).
Conclusions
SHR independently predicts hospitalization HF in STEMI patients post-emergency PCI, demonstrating superior discriminative performance to admission glucose with significant incremental predictive value. SHR may facilitate early risk stratification and management.
Keywords: heart failure, percutaneous coronary intervention, SHR, ST-elevation myocardial infarction, STEMI, stress hyperglycemia ratio
1. Introduction
ST-elevation myocardial infarction (STEMI) represents the most critical type of acute coronary syndrome. Although emergency percutaneous coronary intervention (PCI) and contemporary pharmacological therapies have substantially reduced in-hospital mortality, heart failure (HF) during hospitalization remains one of the most frequent severe complications in STEMI patients, with incidence rates ranging from 14% to 36% (1–3). The occurrence of HF not only prolongs hospital stay and increases healthcare costs but also serves as an independent and powerful predictor of both short-term and long-term mortality (4, 5). Therefore, early identification of high-risk populations for HF development upon admission is of paramount importance for optimizing monitoring levels, guiding hemodynamic management, and improving prognosis.
Stress hyperglycemia refers to transient glucose elevation caused by activation of the sympathetic-adrenal axis and hypothalamic-pituitary-adrenal (HPA) axis during acute illness, leading to massive release of counter-regulatory hormones (cortisol, catecholamines, glucagon, and growth hormone). This phenomenon occurs in patients both with and without a history of diabetes (6, 7). Extensive previous research has confirmed that elevated admission glucose is independently associated with increased infarct size, reduced myocardial salvage, and higher in-hospital mortality and major adverse cardiovascular event (MACE) risk in acute myocardial infarction (AMI) patients (8–10).
However, absolute admission glucose levels as indicators of stress severity have inherent limitations: admission glucose simultaneously reflects both acute stress response and patients’ baseline chronic glycemic status. For diabetic patients with chronically poor glucose control, “elevated” admission glucose may represent only a mild stress response, whereas the same absolute value in non-diabetic patients indicates profound stress (11). This “baseline glucose confounding effect” severely undermines the clinical interpretability of admission glucose as an acute stress biomarker.
To overcome these limitations, Roberts et al. introduced the concept of stress hyperglycemia ratio (SHR) in 2015 (12). SHR is calculated as admission glucose divided by estimated average glucose (eAG), where eAG (mmol/L) = 1.59 × HbA1c (%)−2.59 (13). By standardizing admission glucose to individual chronic glycemic levels, SHR eliminates interference from baseline glucose metabolism and more precisely reflects the relative magnitude of acute stress-induced glucose elevation. In recent years, the prognostic value of SHR has been preliminarily validated in critical care scenarios including acute coronary syndrome, acute ischemic stroke, sepsis, and cardiac surgery (14–17).
Based on this background, the present retrospective cohort study aims to investigate the independent association between SHR and HF risk during hospitalization following emergency PCI in STEMI patients.
2. Methods
2.1. Study design and population
This study was a single-center, retrospective cohort study that consecutively enrolled patients who underwent emergency PCI for STEMI in the Department of Cardiology at Anhui Provincial Hospital from January 2020 to June 2025.Inclusion criteria:① Age ≥18 years; ② Meeting the diagnostic criteria for STEMI according to the Fourth Universal Definition of Myocardial Infarction (18); ③Receiving emergency PCI with a standard indication for STEMI reperfusion therapy, with symptom onset to balloon inflation time strictly limited within 12 h. Exclusion criteria:① Previous chronic heart failure (NYHA Class II–IV); Notably, only patients with symptomatic chronic heart failure were excluded, and patients with asymptomatic left ventricular dysfunction were included in the present study. Left ventricular ejection fraction (LVEF) was fully adjusted in the multivariate regression model to control for the confounding effect of baseline asymptomatic left ventricular systolic dysfunction; ② Severe valvular heart disease; ③ Cardiogenic shock requiring mechanical circulatory support prior to PCI upon admission; ④ Missing admission glucose or HbA1c data; ⑤ HbA1c < 4.0% or >14.0% (extreme values suggesting measurement error); ⑥ Active malignancy or life expectancy <6 months; ⑦ Severe hepatic insufficiency (Child-Pugh Class C).The study protocol was approved by the Ethics Committee of Anhui Provincial Hospital and was conducted in accordance with the Declaration of Helsinki. Given the retrospective nature of the study, individual informed consent was waived.
2.2. Data collection
All data were extracted from the hospital electronic medical records (HIS) system. The recorded variables included the following categories:Demographics and medical history: Age, sex, body mass index (BMI), smoking history, alcohol consumption history, hypertension, diabetes mellitus, dyslipidemia, previous ischemic stroke, previous myocardial infarction, and previous PCI history. Admission vital signs: Heart rate, systolic blood pressure (SBP), and diastolic blood pressure (DBP). Laboratory parameters on admission and during hospitalization: Admission plasma glucose (emergency blood sampling), glycated hemoglobin (HbA1c), peak cardiac troponin I (cTnI), N-terminal pro-B-type natriuretic peptide (NT-proBNP), serum creatinine (Scr), estimated glomerular filtration rate (eGFR, CKD-EPI equation), white blood cell (WBC) count, hemoglobin (Hb),platelet count (PLT),total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides (TG), serum potassium, serum sodium, uric acid (UA),D-dimer, and high-sensitivity C-reactive protein (hs-CRP). Echocardiographic parameters: Left ventricular ejection fraction (LVEF) measured by the first transthoracic echocardiography performed after hospital admission and prior to the occurrence of in-hospital heart failure events during hospitalization. Coronary angiography and procedural variables: Culprit vessel (LAD, LCX, RCA, or LM), number of diseased vessels (single-vessel vs. multivessel disease), door-to-balloon time (D-to-B), and post-procedural TIMI flow grade. All data were extracted from the hospital electronic medical records (HIS) system. The recorded variables included the following categories: Demographics and medical history: Age, sex, body mass index (BMI), smoking history, alcohol consumption history, hypertension, diabetes mellitus, dyslipidemia, previous ischemic stroke, previous myocardial infarction, and previous PCI history. Admission vital signs: Heart rate, systolic blood pressure (SBP), and diastolic blood pressure (DBP). Laboratory parameters on admission and during hospitalization: Admission plasma glucose (emergency blood sampling), glycated hemoglobin (HbA1c), peak cardiac troponin I (cTnI), N-terminal pro-B-type natriuretic peptide (NT-proBNP), serum creatinine (Scr), estimated glomerular filtration rate (eGFR, CKD-EPI equation), white blood cell (WBC) count, hemoglobin (Hb),platelet count (PLT),total cholesterol (TC),low-density lipoprotein cholesterol (LDL-C),high-density lipoprotein cholesterol (HDL-C), triglycerides (TG),serum potassium, serum sodium, uric acid (UA),D-dimer, and high-sensitivity C-reactive protein (hs-CRP). Echocardiographic parameters: Left ventricular ejection fraction (LVEF) measured by the first transthoracic echocardiography performed after hospital admission and prior to the occurrence of in-hospital heart failure events during hospitalization. Coronary angiography and procedural variables: Culprit vessel (LAD, LCX, RCA, or LM), number of diseased vessels (single-vessel vs. multivessel disease), door-to-balloon time (D-to-B), and post-procedural TIMI flow grade. In-hospital medications: Aspirin, P2Y₁₂ receptor inhibitors, statins, β-blockers, ACE inhibitors/angiotensin receptor blockers (ACEI/ARB), and diuretics. Mechanical support devices: Intra-aortic balloon pump (IABP). Mechanical support devices: Intra-aortic balloon pump (IABP).
2.3. Calculation of stress-induced hyperglycemia ratio
SHR was calculated using the formula established by Roberts et al. (12): SHR = admission glucose (mmol/L)/eAG (mmol/L). Estimated average glucose (eAG) was derived from the HbA1c-to-average glucose conversion formula published by Nathan et al. (13): eAG (mmol/L) = 1.59 × HbA1c (%)−2.59. An SHR of 1.0 indicates that admission glucose equals the chronic average glucose level; SHR > 1.0 suggests acute stress-induced hyperglycemia above baseline glucose levels; SHR < 1.0 indicates relative hypoglycemia or acute-phase glucose below baseline levels.
2.4. Definition of primary endpoints
The primary endpoint was definite new-onset or deteriorated in-hospital acute heart failure (AHF) after emergency PCI, which was strictly defined based on standardized clinical manifestations, objective imaging evidence, and standardized therapeutic intervention to reduce phenotypic heterogeneity of study endpoints. Notably, the study endpoint focused on post-PCI incident HF events, excluding persistent heart failure status that existed stably before emergency PCI. The diagnostic criteria must meet at least one of the following objective and confirmatory conditions during hospitalization after PCI: ① New-onset heart failure corresponding to Killip class ≥ II after STEMI, or deterioration of heart failure with Killip grade upgrade compared with pre-PCI baseline status (19); ② Typical acute heart failure signs and symptoms (orthopnea, paroxysmal nocturnal dyspnea, bilateral pulmonary rales involving more than half of the lung fields, obvious jugular venous distension, or peripheral edema) newly occurring after PCI, accompanied by initiation or intravenous intensification of diuretic/vasodilator therapy for HF treatment; ③ Newly emerged objective chest imaging evidence of pulmonary congestion or pulmonary edema consistent with acute heart failure after PCI; ④ New-onset cardiogenic shock secondary to post-PCI myocardial dysfunction (persistent systolic blood pressure <90 mmHg for more than 30 min with end-organ hypoperfusion, requiring inotropic or vasopressor support), which represents severe decompensated heart failure (20, 21).
To ensure endpoint consistency and reduce adjudication bias: Patients with only mild, non-specific dyspnea without objective imaging evidence or standardized HF treatment intervention were excluded from endpoint events. Patients with stable baseline Killip class II before PCI who maintained persistent and unimproved but non-aggravated Killip class II status after PCI without any HF deterioration or new intervention were not defined as endpoint events. All in-hospital HF events were independently adjudicated by two senior attending cardiologists. Any discrepancies in judgment were resolved by discussion and final confirmation by a deputy chief physician specializing in cardiovascular diseases to guarantee the accuracy and homogeneity of endpoint inclusion. Notably, baseline Killip class II–IV heart failure at admission (pre-PCI) that was significantly relieved and improved after PCI and standardized medical treatment, without post-PCI HF deterioration or new HF events, was not adjudicated as an endpoint event, which ensures consistent matching between baseline clinical characteristics and outcome definition.
2.5. Statistical analysis
Normality of continuous variables was assessed using the Shapiro–Wilk test combined with histograms and Q-Q plots. Normally distributed continuous variables were expressed as mean ± standard deviation (x¯ ± SD) and compared using independent-sample t-tests; non-normally distributed variables were expressed as median (interquartile range, IQR) and compared using the Mann–Whitney U test. Categorical variables were expressed as frequencies (percentages) and compared using χ2 tests or Fisher's exact tests. Univariate logistic regression was performed first, with variables showing P < 0.05 entered into the multivariate model. Two models were constructed: Model 1 included SHR as a continuous variable; Model 2 categorized SHR into tertiles [T1 (reference), T2, T3] and tested for trend (P for trend). Multicollinearity was assessed using variance inflation factors (VIF > 5 indicating severe collinearity). Model calibration was comprehensively evaluated by both statistical testing and visual verification. The Hosmer-Lemeshow goodness-of-fit test was performed for quantitative calibration assessment. Additionally, a calibration curve was plotted by stratifying patients into 10 equal risk deciles according to model-predicted in-hospital HF probability, to visually compare the consistency between mean predicted probabilities and actual observed event rates across different risk strata.
Patients were dichotomized into high and low SHR groups based on the median SHR value. Propensity scores were calculated using logistic regression with baseline covariates as independent variables. One-to-one nearest-neighbor matching without replacement was performed using a caliper width of 0.2 times the standard deviation of the logit of the propensity score. Post-matching covariate balance was considered adequate when absolute standardized mean differences (SMD) were <0.10 (22). HF incidence was compared between groups after matching. Using the rcs function from the rms package, multivariate-adjusted logistic regression with restricted cubic splines was fitted using four knots positioned at the 5th, 35th, 65th, and 95th percentiles of SHR distribution. The median SHR was used as the reference value to explore dose-response relationships between SHR and HF risk. Statistical significance was assessed using likelihood ratio tests.
ROC curves were plotted for SHR, admission glucose, HbA1c, and the multivariate logistic model predicting HF, with area under the curve (AUC, C-statistic) calculated. AUC differences between predictors were compared using the DeLong method. The optimal SHR cutoff value was determined using the Youden index, with corresponding sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) reported. A basic clinical model (containing conventional risk factors but excluding SHR) was constructed. After adding SHR to the basic model, incremental predictive value was quantified using likelihood ratio tests (LRT), ΔAUC (DeLong method), category-free net reclassification improvement (NRI), and integrated discrimination improvement (IDI) (23). Akaike information criterion (AIC) and Bayesian information criterion (BIC) were compared between models containing SHR, admission glucose, and the basic model.
Stratified analyses were performed by age (<65 vs. ≥ 65 years), sex, diabetes status, Killip class (I–II vs. III–IV), LVEF (<50% vs. ≥ 50%), number of diseased vessels (single-vessel vs. multivessel disease), and renal function (eGFR <60 vs. ≥ 60 mL/min/1.73 m2). Interactions between SHR and stratification variables were tested using multiplicative interaction terms. Sensitivity analyses included: ① Separate analyses in diabetic and non-diabetic subgroups; ② Repeating the primary analysis after excluding Killip class IV patients; ③ Repeating the primary analysis after excluding patients with eGFR <30 mL/min/1.73 m2; ④ Comparing AIC and BIC between models containing SHR versus admission glucose. All statistical analyses were performed using R software (version 4.3.1). Two-sided P < 0.05 was considered statistically significant.
3. Results
3.1. Baseline characteristics
A total of 1,068 patients were included in this study. All baseline clinical characteristics analyzed in this section were collected at hospital admission and before emergency PCI. In-hospital medication data were collected during post-PCI hospitalization. Among them, 158 patients (14.8%) experienced new-onset or deteriorated heart failure during hospitalization after emergency PCI, while 910 patients (85.2%) did not develop post-PCI HF events. Stratified breakdown of the composite in-hospital HF endpoints among 158 event patients was further performed: 117 patients presented with mild-to-moderate HF events corresponding to new-onset or deteriorated Killip class II–III heart failure with clinical pulmonary congestion and required intensified diuretic or vasodilator therapy; 41 patients developed severe HF events, including Killip class IV heart failure and cardiogenic shock requiring vasopressor/inotropic support or mechanical circulatory assistance. Severe HF events accounted for 25.9% of all in-hospital HF cases, indicating that elevated SHR was predictive of both mild and severe post-PCI HF complications across a broad clinical severity spectrum. Baseline characteristics and in-hospital treatment information stratified by in-hospital HF status are shown in Table 1. Compared with the non-HF group, patients in the HF group were older (67.3 ± 11.5 vs. 60.8 ± 12.2 years; P < 0.001), had a higher proportion of females (31.6% vs. 21.8%; P = 0.008), and higher prevalence of hypertension (62.0% vs. 48.1%; P = 0.001) and diabetes mellitus (38.0% vs. 24.2%; P < 0.001). At admission, the HF group had faster heart rate (88.5 ± 18.2 vs. 76.3 ± 14.5 bpm), lower systolic blood pressure (118.5 ± 24.3 vs. 128.6 ± 21.5 mmHg), and lower LVEF (42.8 ± 9.5% vs. 55.6 ± 8.2%). A higher proportion had LAD as the culprit vessel (52.5% vs. 37.8%), multivessel disease was more common (58.2% vs. 41.5%), and D-to-B time was longer [median 120 (82–178) vs. 95 (66–138) min].Regarding the core exposure variables, the HF group had significantly higher admission glucose [median 10.8 (8.2–14.5) vs. 7.5 (5.9–9.6) mmol/L; P < 0.001], higher HbA1c (6.5 ± 1.6% vs. 6.1 ± 1.2%; P = 0.002), and markedly elevated SHR [median 1.52 (1.25–1.88) vs. 1.10 (0.92–1.30); P < 0.001]. The HF group also showed significantly higher levels of peak cTnI, NT-proBNP, WBC, hs-CRP, D-dimer, and UA, while eGFR, Hb, HDL-C, and serum sodium were lower (all P < 0.05).
Table 1.
Baseline characteristics stratified by heart failure status during hospitalization.
| Variables | Total (n = 1,068) | HF group (n = 158) | Non-HF group (n = 910) | P-value |
|---|---|---|---|---|
| Age (years) | 61.8 ± 12.2 | 67.3 ± 11.5 | 60.8 ± 12.2 | < 0.001 |
| Male, n(%) | 820 (76.8) | 108 (68.4) | 712 (78.2) | 0.008 |
| BMI (kg/m2) | 24.7 ± 3.2 | 25.1 ± 3.4 | 24.6 ± 3.2 | 0.085 |
| Smoke, n(%) | 548 (51.3) | 76 (48.1) | 472 (51.9) | 0.383 |
| Alcohol, n(%) | 258 (24.2) | 35 (22.2) | 223 (24.5) | 0.517 |
| Hypertension, n(%) | 536 (50.2) | 98 (62.0) | 438 (48.1) | 0.001 |
| Diabetes, n(%) | 280 (26.2) | 60 (38.0) | 220 (24.2) | < 0.001 |
| Dyslipidemia, n(%) | 328 (30.7) | 55 (34.8) | 273 (30.0) | 0.223 |
| Previous stroke, n(%) | 81 (7.6) | 19 (12.0) | 62 (6.8) | 0.023 |
| Previous PCI, n(%) | 90 (8.4) | 16 (10.1) | 74 (8.1) | 0.409 |
| Previous MI, n(%) | 60 (5.6) | 14 (8.9) | 46 (5.1) | 0.051 |
| Heart rate (bpm) | 78.1 ± 15.5 | 88.5 ± 18.2 | 76.3 ± 14.5 | < 0.001 |
| SBP (mmHg) | 127.1 ± 22.2 | 118.5 ± 24.3 | 128.6 ± 21.5 | < 0.001 |
| DBP (mmHg) | 77.6 ± 12.5 | 72.3 ± 13.8 | 78.5 ± 12.2 | < 0.001 |
| Killip grade, n(%) | < 0.001 | |||
| Ⅰ | 702 (65.7) | 47 (29.7) | 655 (72.0) | |
| Ⅱ | 222 (20.8) | 56 (35.4) | 166 (18.2) | |
| Ⅲ | 99 (9.3) | 35 (22.2) | 64 (7.0) | |
| Ⅳ | 45 (4.2) | 20 (12.7) | 25 (2.7) | |
| LVEF (%) | 53.7 ± 9.6 | 42.8 ± 9.5 | 55.6 ± 8.2 | < 0.001 |
| Admission blood glucose (mmol/L)a | 7.8 (6.1–10.5) | 10.8 (8.2–14.5) | 7.5 (5.9–9.6) | < 0.001 |
| HbA1c (%) | 6.2 ± 1.3 | 6.5 ± 1.6 | 6.1 ± 1.2 | 0.002 |
| SHRa | 1.14 (0.94–1.40) | 1.52 (1.25–1.88) | 1.10 (0.92–1.30) | < 0.001 |
| Peak cTnI (ng/mL)a | 15.2 (7.8–30.5) | 35.6 (18.2–58.3) | 12.8 (6.5–25.6) | < 0.001 |
| NT-proBNP (pg/mL)a | 568 (185–2 150) | 3 520 (1 680–6 850) | 448 (152–1 280) | < 0.001 |
| Scr (μmol/L)a | 80.2 (66.5–98.3) | 95.5 (72.8–125.6) | 78.2 (65.2–94.5) | < 0.001 |
| eGFR (mL/min/1.73 m2) | 85.6 ± 23.8 | 72.5 ± 26.2 | 87.9 ± 22.5 | < 0.001 |
| WBC (×10⁹/L)a | 10.0 (7.8–12.5) | 11.5 (9.2–14.8) | 9.8 (7.6–12.2) | < 0.001 |
| Hb (g/L) | 136.8 ± 17.0 | 128.5 ± 18.3 | 138.2 ± 16.5 | < 0.001 |
| PLT (×10⁹/L) | 215 ± 62 | 212 ± 65 | 216 ± 61 | 0.452 |
| TC (mmol/L) | 4.65 ± 1.10 | 4.72 ± 1.15 | 4.64 ± 1.09 | 0.408 |
| LDL-C (mmol/L) | 2.81 ± 0.89 | 2.95 ± 0.92 | 2.78 ± 0.88 | 0.028 |
| HDL-C (mmol/L) | 1.10 ± 0.30 | 1.02 ± 0.28 | 1.12 ± 0.30 | < 0.001 |
| TG (mmol/L)a | 1.55 (1.02–2.28) | 1.58 (1.05–2.35) | 1.54 (1.02–2.25) | 0.562 |
| Blood potassium (mmol/L) | 4.05 ± 0.50 | 4.08 ± 0.52 | 4.04 ± 0.49 | 0.348 |
| Blood sodium (mmol/L) | 140.0 ± 3.3 | 138.5 ± 3.8 | 140.2 ± 3.2 | < 0.001 |
| UA (μmol/L) | 347 ± 97 | 385 ± 105 | 340 ± 95 | < 0.001 |
| hs-CRP (mg/L)a | 8.5 (3.8–18.5) | 18.2 (9.5–35.6) | 7.8 (3.5–16.2) | < 0.001 |
| D-dimer (mg/L)a | 0.85 (0.42–1.68) | 1.82 (0.95–3.28) | 0.78 (0.38–1.52) | < 0.001 |
| Culprit vessel, n(%) | 0.001 | |||
| LAD | 427 (40.0) | 83 (52.5) | 344 (37.8) | |
| LCX | 152 (14.2) | 19 (12.0) | 133 (14.6) | |
| RCA | 436 (40.8) | 47 (29.7) | 389 (42.7) | |
| LM | 53 (5.0) | 9 (5.7) | 44 (4.8) | |
| Multivessel disease, n(%) | 470 (44.0) | 92 (58.2) | 378 (41.5) | < 0.001 |
| D-to-B time (min)a | 98 (68–145) | 120 (82–178) | 95 (66–138) | < 0.001 |
| TIMI 3 POST PCI, n(%) | 926 (86.7) | 119 (75.3) | 807 (88.7) | < 0.001 |
| Aspirin, n(%) | 1 035 (96.9) | 150 (94.9) | 885 (97.3) | 0.115 |
| P2Y₁₂ inhibitor, n(%) | 1 025 (96.0) | 148 (93.7) | 877 (96.4) | 0.105 |
| Statins, n(%) | 1 013 (94.9) | 145 (91.8) | 868 (95.4) | 0.068 |
| Beta blockers, n(%) | 859 (80.4) | 114 (72.2) | 745 (81.9) | 0.005 |
| ACEI/ARB, n(%) | 672 (62.9) | 107 (67.7) | 565 (62.1) | 0.168 |
| Diuretics, n(%) | 278(26.0) | 114(72.2) | 164(18.0) | < 0.001 |
| IABP, n(%) | 51(4.8) | 24(15.2) | 27(3.0) | < 0.001 |
Non-normally distributed variables are expressed as median (IQR).
BMI, Body Mass Index; SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure; LVEF, Left Ventricular Ejection Fraction; LAD, Left Anterior Descending artery; LCX, Left Circumflex artery; RCA, Right Coronary Artery; LM, Left Main Coronary Artery; D-to-B, Door-to-Balloon Time; TIMI, Thrombolysis In Myocardial Infarction; IABP, Intra-Aortic Balloon Pump.
All baseline data (demographics, medical history, admission vital signs, laboratory indexes, and angiographic characteristics) were collected at hospital admission and before emergency PCI. All in-hospital medication data and IABP application status were recorded during hospitalization after emergency PCI. In the Non-HF group, patients with baseline Killip class II–IV had acute heart failure symptoms caused by STEMI at admission; all these patients achieved significant symptomatic improvement and Killip class regression after PCI and in-hospital treatment, without post-procedural new-onset heart failure or aggravated HF events. The 18% diuretic use rate in the Non-HF group was attributed to low-dose prophylactic diuretic therapy for mild post-STEMI volume overload or mild pulmonary congestion; these patients did not receive intensified HF treatment and did not meet the diagnostic criteria for study-defined in-hospital acute heart failure endpoints.
3.2. SHR tertiles and incidence of heart failure
Patients were equally divided into three tertiles based on SHR distribution: T1 (SHR: 0.30–0.98, n = 356), T2 (SHR: 0.99–1.28, n = 356), and T3 (SHR: 1.29–5.00, n = 356). The incidence of in-hospital HF increased progressively across SHR tertiles: T1 = 5.9% (21/356), T2 = 12.4% (44/356), T3 = 26.1% (93/356) (χ2 for trend = 52.8, P for trend < 0.001; Figure 1).
Figure 1.
Trends In in-hospital heart failure incidence across three SHR tertile subgroups. X-axis labels represent the absolute SHR range for each tertile: T1 (0.30–0.98), T2 (0.99–1.28), T3 (1.29–5.00).
3.3. Univariate and multivariate logistic regression analysis
In univariate analysis, each 1-unit increase in SHR was associated with a crude OR of 4.85 (95% CI 3.52–6.69; P < 0.001) for HF occurrence, representing the most significant effect size among all candidate predictors (Table 2). Variables with P < 0.05 in univariate analysis were entered into the multivariate logistic regression model. All included variables had VIF < 3.5, indicating no severe multicollinearity. The Hosmer-Lemeshow goodness-of-fit test yielded χ2 = 8.45, P = 0.391, indicating acceptable statistical calibration (Table 3). Further visual calibration analysis via risk decile stratification demonstrated excellent agreement between predicted and observed HF probabilities (Figure 2). The calibration curve showed no systematic overestimation or underestimation of in-hospital HF risk across low-, medium-, and high-risk patient subgroups, verifying the robust fitting performance of the multivariate predictive model. After multivariate adjustment, the independent predictive effect of SHR remained highly significant despite some attenuation from the crude OR: each 1-unit increase yielded an adjusted OR of 2.31 (95% CI 1.68–3.18; P < 0.001). In the tertile analysis, using T1 as the reference, T3 showed an adjusted OR of 2.86 (95% CI 1.62–5.05; P < 0.001), with P for trend < 0.001, indicating a significant “dose-response relationship” between SHR and HF risk. Within the same model, Killip class (adjusted OR=1.92 per grade) and LVEF (adjusted OR=0.92 per %) were also independent predictors, while hypertension and diabetes mellitus lost statistical significance after inclusion of SHR.
Table 2.
Univariate logistic regression analysis of heart failure during hospitalization.
| Variables | OR | 95% CI | P-value |
|---|---|---|---|
| Age (per additional year) | 1.04 | 1.03–1.06 | < 0.001 |
| Male | 0.61 | 0.42–0.87 | 0.008 |
| BMI (per additional 1 kg/m2) | 1.05 | 0.99–1.11 | 0.085 |
| Smoke | 0.86 | 0.61–1.20 | 0.383 |
| Hypertension | 1.77 | 1.26–2.49 | 0.001 |
| Diabetes | 1.92 | 1.36–2.72 | < 0.001 |
| Dyslipidemia | 1.25 | 0.87–1.78 | 0.223 |
| Previous strok | 1.87 | 1.08–3.22 | 0.025 |
| Previous MI | 1.83 | 0.98–3.40 | 0.057 |
| Killip Grade (per additional 1 Grade) | 2.85 | 2.38–3.41 | < 0.001 |
| Heart Rate (per additional 1 bpm) | 1.04 | 1.03–1.05 | < 0.001 |
| SBP (per additional 1 mmHg) | 0.98 | 0.97–0.99 | < 0.001 |
| LVEF (per additional 1%) | 0.87 | 0.85–0.89 | < 0.001 |
| Multivessel disease | 1.96 | 1.40–2.75 | < 0.001 |
| D-to-BTime (per additional 1 min) | 1.003 | 1.001–1.005 | 0.003 |
| Admission blood glucose (per additional 1 mmol/L) | 1.18 | 1.14–1.22 | < 0.001 |
| SHR (per unit increase) | 4.85 | 3.52–6.69 | < 0.001 |
| Peak cTnI (per additional 1 ng/mL) | 1.02 | 1.01–1.02 | < 0.001 |
| NT-proBNP (per additional 100 pg/mL) | 1.03 | 1.02–1.04 | < 0.001 |
| eGFR (per additional 1 mL/min/1.73 m2) | 0.97 | 0.96–0.98 | < 0.001 |
| WBC (per additional 1 × 10⁹/L) | 1.10 | 1.05–1.15 | < 0.001 |
| Hb (per additional 1 g/L) | 0.97 | 0.96–0.98 | < 0.001 |
| LDL-C (per additional 1 mmol/L) | 1.22 | 1.02–1.46 | 0.028 |
| HDL-C (per additional 1 mmol/L) | 0.38 | 0.19–0.76 | 0.006 |
| Blood sodium (per additional 1 mmol/L) | 0.92 | 0.87–0.96 | < 0.001 |
| UA (per additional 1 μmol/L) | 1.004 | 1.002–1.006 | < 0.001 |
| D-dimer (per additional 1 mg/L) | 1.18 | 1.10–1.27 | < 0.001 |
| hs-CRP (per additional 1 mg/L) | 1.02 | 1.01–1.03 | < 0.001 |
Table 3.
Multivariate logistic regression analysis of heart failure during hospitalization.
| Variables | Model 1:SHR continuous variables | P-value | Model 2:SHR tertiles | P-value |
|---|---|---|---|---|
| Adjusted OR(95% CI) | Adjusted OR(95% CI) | |||
| SHR (per unit) | 2.31 (1.68–3.18) | < 0.001 | — | — |
| SHR T1(References) | — | — | 1.00(References) | — |
| SHR T2 vs. T1 | — | — | 1.62 (0.89–2.95) | 0.112 |
| SHR T3 vs. T1 | — | — | 2.86 (1.62–5.05) | < 0.001 |
| Trend P | — | — | — | < 0.001 |
| Age (per year) | 1.02 (1.00–1.04) | 0.038 | 1.02 (1.00–1.04) | 0.042 |
| Male | 0.78 (0.48–1.26) | 0.312 | 0.76 (0.47–1.24) | 0.276 |
| Hypertension | 1.18 (0.78–1.80) | 0.436 | 1.20 (0.79–1.83) | 0.398 |
| Diabetes | 1.35 (0.86–2.12) | 0.192 | 1.38 (0.88–2.17) | 0.165 |
| Killip grade (per grade) | 1.92 (1.53–2.41) | < 0.001 | 1.95 (1.55–2.45) | < 0.001 |
| Heart rate (per bpm) | 1.01 (1.00–1.03) | 0.068 | 1.01 (1.00–1.03) | 0.072 |
| SBP (per mmHg) | 0.99 (0.98–1.00) | 0.185 | 0.99 (0.98–1.00) | 0.192 |
| LVEF (per %) | 0.92 (0.90–0.94) | < 0.001 | 0.92 (0.90–0.94) | < 0.001 |
| Multivessel disease | 1.28 (0.83–1.97) | 0.262 | 1.30 (0.85–2.00) | 0.228 |
| D-to-B time (per min) | 1.001 (0.999–1.003) | 0.225 | 1.001 (0.999–1.003) | 0.218 |
| Peak cTnI (per ng/mL) | 1.01 (1.00–1.01) | 0.012 | 1.01 (1.00–1.01) | 0.010 |
| NT-proBNP (per 100 pg/mL) | 1.01 (1.00–1.02) | 0.005 | 1.01 (1.00–1.02) | 0.006 |
| eGFR (per mL/min/1.73 m2) | 0.99 (0.98–1.00) | 0.128 | 0.99 (0.98–1.00) | 0.135 |
| WBC (per × 10⁹/L) | 1.03 (0.97–1.09) | 0.318 | 1.03 (0.97–1.09) | 0.325 |
| Hb (per g/L) | 0.99 (0.98–1.00) | 0.218 | 0.99 (0.98–1.00) | 0.225 |
| HDL-C (per mmol/L) | 0.62 (0.28–1.38) | 0.245 | 0.60 (0.27–1.34) | 0.218 |
| Blood sodium (per mmol/L) | 0.95 (0.90–1.01) | 0.082 | 0.95 (0.90–1.01) | 0.085 |
| hs-CRP (per mg/L) | 1.01 (1.00–1.02) | 0.068 | 1.01(1.00–1.02) | 0.072 |
Figure 2.
Calibration Curve of the multivariate logistic regression model for predicting in-hospital heart failure. The x-axis represents the mean predicted in-hospital HF probability stratified by ten equal risk deciles; the y-axis represents the actual observed in-hospital HF incidence in each risk decile. The dashed diagonal line indicates the ideal perfect calibration line where predicted probability equals observed probability. The scatter points and fitted curve show favorable agreement between predicted and observed event rates across all risk strata, confirming excellent calibration of the SHR-based predictive model.
3.4. Restrictive cubic spline analysis
To explore the dose-response relationship between SHR and in-hospital heart failure risk, restricted cubic spline (RCS) regression analysis was performed. As shown in Figure 3, the RCS curve demonstrated a monotonically increasing trend: as SHR increased from 0.5 to 3.5, the adjusted OR progressively rose from below 1.0, indicating that higher SHR values were associated with greater risk of in-hospital heart failure. The overall association between SHR and heart failure was highly statistically significant (P overall < 0.001), while the test for nonlinearity was not significant (P nonlinear = 0.452), suggesting a linear dose-response relationship whereby each 1-unit increase in SHR was associated with a relatively constant proportional increase in heart failure risk, with no apparent threshold or plateau effects observed. Notably, when SHR exceeded approximately 1.2–1.3, the lower limit of the 95% confidence interval began to exceed the reference line (OR = 1.0), which closely aligned with the optimal cutoff value of 1.27 determined by the Youden index in ROC analysis. Consistent with visual observation, the confidence bands widened substantially at the extreme upper range of the SHR distribution (SHR > 2.5), which was attributed to data sparsity and limited sample size in the extreme high SHR subgroup.
Figure 3.
Dose-response Relationship between SHR and risk of in-hospital heart failure.
3.5. Propensity score matching analysis
Patients were divided into high SHR (n = 534) and low SHR (n = 534) groups based on the median SHR value (1.14). Before matching, the two groups differed significantly in multiple baseline characteristics including age, diabetes mellitus, Killip class, and LVEF. After 1:1 nearest-neighbor PSM, 432 pairs (864 patients) were successfully matched. Post-matching, all covariates achieved absolute SMD < 0.10, indicating well-balanced covariates. After matching, the high SHR group had twice the incidence of in-hospital HF compared to the low SHR group (20.4% vs. 10.2%; OR=2.26, 95% CI 1.52–3.36; P < 0.001). Additionally, the high SHR group demonstrated significantly higher in-hospital mortality (3.7% vs. 1.2%; OR=3.28; P = 0.018) and longer length of stay [median 9 (7–13) vs. 7 (6–10) days; P < 0.001] compared to the low SHR group. The PSM results were highly consistent with the multivariate logistic regression findings, confirming the robustness of the association between SHR and HF (Table 4).
Table 4.
Comparison of outcomes between the high SHR group and the Low SHR group after propensity score matching.
| Outcome | High SHR Group (n = 432) | Low SHR Group(n = 432) | OR(95% CI) | P-Value |
|---|---|---|---|---|
| HF during hospitalization, n(%) | 88 (20.4) | 44 (10.2) | 2.26 (1.52–3.36) | < 0.001 |
| In-hospital death, n(%) | 16 (3.7) | 5 (1.2) | 3.28 (1.19–9.03) | 0.018 |
| Days in hospital (days)a | 9 (7–13) | 7 (6–10) | — | < 0.001 |
Non-normally distributed variables are expressed as median (IQR).
3.6. ROC curve analysis with optimal cutoff
The AUC of SHR for predicting in-hospital HF was 0.748 (95% CI 0.711–0.785), which was significantly superior to admission glucose (AUC = 0.728; DeLong P = 0.031) and HbA1c (AUC = 0.548; P < 0.001). The multivariate model integrating SHR with clinical covariates achieved a high AUC of 0.812 (95% CI 0.792–0.832). The optimal SHR cutoff determined by the Youden index was 1.27, corresponding to a sensitivity of 72.8%, specificity of 67.5%, PPV of 28.0%, and NPV of 93.4%. The high NPV (93.4%) indicates that SHR < 1.27 can effectively rule out in-hospital HF occurrence (Table 5). As shown in Figure 4, the ROC curve of SHR consistently positioned above and to the left of the admission glucose curve, while the HbA1c curve closely followed the diagonal reference line.
Table 5.
ROC curve analysis for predicting heart failure during hospitalization.
| Predictors | AUC (95% CI) | Cut-Off | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) | vs. SHR DeLong P |
|---|---|---|---|---|---|---|---|
| SHR | 0.748 (0.711–0.785) | 1.27 | 72.8 | 67.5 | 28.0 | 93.4 | — |
| Admission blood glucose | 0.728 (0.690–0.766) | 9.2 mmol/L | 68.4 | 66.8 | 26.4 | 92.4 | 0.031 |
| HbA1c | 0.548 (0.503–0.593) | 6.2% | 55.1 | 52.8 | 16.9 | 87.2 | < 0.001 |
| Multivariate model | 0.812 (0.792–0.832) | — | 83.5 | 85.2 | 49.5 | 96.8 | — |
Figure 4.
ROC Curve analysis.
3.7. Incremental predictive value
After incorporating SHR into the baseline model containing conventional risk factors: AUC improved from 0.898 to 0.912 (ΔAUC = + 0.014; P = 0.008); likelihood ratio test χ2 = 18.6, P < 0.001; continuous NRI = 0.318 (P < 0.001), with event NRI = + 0.224 indicating that 22.4% of HF patients were correctly reclassified to higher risk categories, and non-event NRI = + 0.094 indicating that 9.4% of non-HF patients were correctly reclassified to lower risk categories; IDI = 0.035 (P < 0.001). The SHR-containing model demonstrated lower AIC (645.2) and BIC (724.8) compared to both the baseline model (662.8/742.4) and the admission glucose model (658.7/738.3), supporting superior model fit for SHR (Table 6).
Table 6.
Incremental predictive value of SHR beyond the basic clinical model.
| Assessment indicators | Value | 95%CI/P-value |
|---|---|---|
| Base model AUC | 0.898 | 0.876–0.920 |
| Base model + SHR AUC | 0.912 | 0.892–0.932 |
| ΔAUC | +0.014 | DeLong P = 0.008 |
| Likelihood ratio test χ2 | 18.6 | P < 0.001 |
| Categoryless NRI | 0.318 | P < 0.001 |
| NRI(event) | +0.224 | — |
| NRI(non-event) | +0.094 | — |
| IDI | 0.035 | P < 0.001 |
| Base model AIC | 662.8 | — |
| Base model + SHR AIC | 645.2 | — |
| Base model + admission blood glucose AIC | 658.7 | — |
| Base model BIC | 742.4 | — |
| Base model + SHR BIC | 724.8 | — |
| Base model + admission blood glucose BIC | 738.3 | — |
The basic clinical model was adjusted for all conventional prognostic risk factors, including age, sex, body mass index, smoking history, alcohol consumption, hypertension, diabetes mellitus, dyslipidemia, previous ischemic stroke, previous myocardial infarction, previous PCI history, heart rate, systolic blood pressure, diastolic blood pressure, peak cardiac troponin I, NT-proBNP, eGFR, white blood cell count, high-sensitivity C-reactive protein, D-dimer, left ventricular ejection fraction, culprit vessel type, multivessel disease, door-to-balloon time, and post-procedural TIMI flow grade. This baseline model did not include SHR or admission glucose.
3.8. Subgroup analysis and sensitivity analysis
As shown in the forest plot (Figure 5), the association between elevated SHR and increased risk of in-hospital HF was consistent in direction across all predefined subgroups (all ORs > 1). All interaction P values were > 0.05, indicating that the predictive effect of SHR was not significantly modified by age, sex, LVEF level, extent of coronary disease, or renal function status. Notably, there was a borderline interaction trend for diabetes status (interaction P = 0.068): the effect size of SHR was numerically greater in non-diabetic patients (OR=2.65) than in diabetic patients (OR=1.78), though the 95% CIs of both groups overlapped (Table 7). Across various sensitivity analyses, the independent predictive effect of SHR remained robust: adjusted ORs fluctuated between 1.85–2.58, with consistent direction and significance (Table 8).
Figure 5.
Forest Plot for subgroup analyses.
Table 7.
Subgroup analysis: association between high SHR (≥median) and heart failure during hospitalization.
| Subgroups | n | Number of events | OR(95% CI) | P-value | Interaction p-value |
|---|---|---|---|---|---|
| Age | 0.385 | ||||
| <65 | 592 | 62 | 2.45 (1.42–4.22) | 0.001 | |
| ≥65 | 476 | 96 | 2.12 (1.32–3.40) | 0.002 | |
| Gender | 0.528 | ||||
| Male | 820 | 108 | 2.18 (1.42–3.35) | < 0.001 | |
| Female | 248 | 50 | 2.55 (1.32–4.92) | 0.005 | |
| Diabetes | 0.068 | ||||
| No | 780 | 98 | 2.65 (1.72–4.08) | < 0.001 | |
| Yes | 288 | 60 | 1.78 (1.08–2.93) | 0.024 | |
| Killip Grade | 0.312 | ||||
| Ⅰ–Ⅱ | 924 | 103 | 2.38 (1.55–3.65) | < 0.001 | |
| Ⅲ–Ⅳ | 144 | 55 | 1.95 (0.98–3.88) | 0.058 | |
| LVEF | 0.425 | ||||
| ≥50% | 685 | 52 | 2.52 (1.38–4.60) | 0.003 | |
| <50% | 383 | 106 | 2.08 (1.28–3.38) | 0.003 | |
| Vascular Lesions | 0.562 | ||||
| Single branch | 598 | 66 | 2.35 (1.38–4.00) | 0.002 | |
| Multiple branches | 470 | 92 | 2.18 (1.35–3.52) | 0.002 | |
| eGFR | 0.218 | ||||
| ≥60 | 852 | 108 | 2.42 (1.58–3.71) | < 0.001 | |
| <60 | 216 | 50 | 1.85 (0.98–3.48) | 0.056 |
Table 8.
Sensitivity analysis.
| Analysis | n | Adjusted OR (95% CI) | P-value |
|---|---|---|---|
| Main analysis (full cohort) | 1 068 | 2.31 (1.68–3.18) | < 0.001 |
| Non-diabetic subgroup | 780 | 2.58 (1.72–3.87) | < 0.001 |
| Diabetic subgroup | 288 | 1.85 (1.12–3.05) | 0.016 |
| Exclude patients with Killip IV | 1 023 | 2.25 (1.62–3.12) | < 0.001 |
| Exclude eGFR <30 | 1 042 | 2.28 (1.65–3.15) | < 0.001 |
| PSM matching cohort | 864 | 2.26 (1.52–3.36) | < 0.001 |
4. Discussion
In this single-center cross-sectional study including 1,068 STEMI patients undergoing emergent PCI, we systematically evaluated the predictive value of the stress hyperglycemia ratio (SHR) for in-hospital heart failure. The main findings were as follows: SHR was a strong independent predictor of in-hospital HF. After multivariate adjustment for covariates, each 1-unit increase in SHR was associated with a 2.31-fold increase in HF risk (adjusted OR = 2.31, 95% CI 1.68–3.18). The highest SHR tertile (T3) demonstrated nearly 3-fold higher HF risk compared to the lowest tertile (T1) (adjusted OR = 2.86), with a significant dose-response trend (P for trend < 0.001). A linear dose-response relationship existed between SHR and HF risk. RCS analysis revealed that each 1-unit increase in SHR was associated with a relatively constant proportional increase in heart failure risk, with no apparent threshold or plateau effects observed. The SHR-HF association demonstrated robustness through propensity score matching validation. After matching, the high SHR group maintained more than twice the HF incidence of the low SHR group (20.4% vs. 10.2%). SHR showed superior discriminatory performance compared to conventional glycemic indices. The AUC for SHR (0.748) was significantly higher than that of admission glucose (0.728; P = 0.031) and HbA1c (0.548; P < 0.001). SHR provided significant incremental predictive value. When added to the baseline model, NRI = 0.318 and IDI = 0.035 (both P < 0.001), with superior AIC and BIC compared to the admission glucose model. The predictive effect of SHR remained consistent across all predefined subgroups, unmodified by age, sex, diabetes status, LVEF level, extent of coronary disease, or renal function status.
Our findings are consistent with and extend recent literature on the prognostic value of SHR in acute cardiovascular events. Yang et al. (14) reported that elevated SHR was an independent predictor of MACE during a median follow-up of 2.1 years in 1,262 AMI patients undergoing PCI (highest vs. lowest quartile HR = 1.89). Xu et al. (15) found that SHR could predict no-reflow phenomenon and in-hospital mortality in STEMI patients. Gao et al. (16), based on MIMIC-IV database analysis, demonstrated that SHR was superior to admission glucose in predicting 30-day and 1-year mortality after AMI.
However, most previous studies focused on composite MACE endpoints or long-term mortality, with few specifically investigating in-hospital HF as an independent primary endpoint. This study is the first to systematically examine in-hospital HF as the primary endpoint, integrating multivariate regression, PSM, RCS, and incremental predictive value analyses to provide the most comprehensive evidence to date regarding the SHR-in-hospital HF association. The observed in-hospital HF incidence (14.8%) in our study was consistent with contemporary PCI-era registry studies. Bahit et al. reported a 13.8% in-hospital HF incidence in the STEMI subgroup of the PLATO trial, while Desta et al. reported 16.2% in the SWEDEHEART registry study. Our data fall within this range, suggesting representativeness of our study population.
The association between SHR and post-STEMI HF has a solid pathophysiological foundation involving multiple interconnected mechanisms: Acute hyperglycemia induces cardiomyocyte injury through multiple pathways: ① excessive superoxide anion (O₂⁻) production by the mitochondrial electron transport chain, initiating oxidative stress cascades (24); ② activation of the polyol and hexosamine pathways, leading to increased protein O-GlcNAc modifications that interfere with calcium-regulating protein function (25); ③ promotion of advanced glycation end product (AGE) formation and their receptor (RAGE) signaling, triggering pro-fibrotic and pro-apoptotic gene expression (26). In the context of ischemia-reperfusion (I/R), hyperglycemia exacerbates reperfusion injury by attenuating ischemic preconditioning and lowering the threshold for mitochondrial permeability transition pore (mPTP) opening (27). Stress hyperglycemia impairs microcirculatory perfusion by reducing nitric oxide (NO) bioavailability, increasing endothelin-1 (ET-1) expression, and upregulating vascular cell adhesion molecule-1 (VCAM-1) (28). Even after successful epicardial vessel recanalization, microvascular obstruction (MVO) can persist, leading to expansion of the effective infarct area and ventricular remodeling.
Elevated SHR essentially reflects the severity of acute stress response: excessive sympatho-adrenal system activation results in dramatic increases in circulating catecholamine concentrations, which directly increase myocardial oxygen consumption and arrhythmia risk while causing myocardial stunning through β1-adrenergic receptor-mediated calcium overload (29). HPA axis activation leads to cortisol surges, further exacerbating insulin resistance and hyperglycemia. Furthermore, acute hyperglycemia can directly activate inflammatory signaling pathways including NF-κB (30), Toll-like receptor-4 (TLR-4), and the NLRP3 inflammasome (31), driving the cascading release of interleukin-1β (IL-1β), interleukin-6 (IL-6), and tumor necrosis factor-α (TNF-α). These pro-inflammatory mediators not only aggravate acute-phase myocardial injury but also promote chronic fibrosis and adverse remodeling, forming the pathological bridge from acute injury to heart failure.
Hyperglycemia enhances platelet GPIIb/IIIa receptor expression and platelet aggregation activity (32), upregulates tissue factor (TF) expression (33), and inhibits fibrinolysis by elevating plasminogen activator inhibitor-1 (PAI-1) levels (34). These prothrombotic effects may promote microvascular thrombosis, in-stent thrombosis, and infarct expansion. The normal heart primarily relies on fatty acid β-oxidation for ATP generation, but shifts toward more “oxygen-efficient” aerobic and anaerobic glucose metabolism under ischemic conditions (35). Acute insulin resistance (a core feature of elevated SHR) severely impairs glucose uptake capacity in ischemic myocardium by inhibiting glucose transporter 4 (GLUT-4) translocation to the cell membrane, forcing the myocardium to be “deprived” of this critical substrate precisely when glucose is most needed this “metabolic inflexibility” represents an important mechanism leading to further deterioration of contractile function and pump failure (36).
Notably, the widened confidence intervals in the high extreme SHR range in RCS analysis deserve explicit interpretation. The substantial broadening of confidence bands for SHR values greater than 2.5 is caused by data sparsity in the upper tail of the SHR distribution, where only a small number of patients presented with extremely elevated stress hyperglycemia. This sparse data distribution increases statistical sampling uncertainty and reduces the model's quantitative predictive reliability for extreme high-risk outliers. Nevertheless, the tightly narrowed and stable confidence intervals across the main SHR distribution range (0.5–2.5) firmly support the robust linear dose-response relationship between SHR and in-hospital HF risk in the vast majority of STEMI patients. This limitation only affects extreme outliers rather than the overall core conclusion of the study.
The AUC of SHR (0.748) was significantly superior to that of admission glucose (0.728; P = 0.031) and HbA1c (0.548; P < 0.001). Notably, as a derived index calculated from admission glucose and HbA1c, SHR has inherent interpretative limitations in ROC analysis, and its predictive efficacy cannot be completely decoupled from the two original glycemic indicators. The superior discriminative ability of SHR does not originate from novel independent biomarker information, but from the individualized standardization of admission glucose based on patients’ chronic glycemic status, which effectively removes the confounding effect of baseline glucose metabolism and improves the accuracy of acute stress hyperglycemia evaluation. Consistent with our ROC findings, subsequent incremental predictive value analyses further confirmed that SHR provided significant NRI and IDI for HF risk stratification compared with raw admission glucose, with better model fitness (lower AIC and BIC). These results collectively demonstrate that SHR is a clinically optimized glycemic index rather than a simply mathematically derived variable, with unique practical value for identifying post-PCI HF high-risk patients that cannot be replaced by absolute admission glucose levels.
This study has the following limitations that warrant consideration: First, this single-center retrospective study had inherent data limitations. Detailed electrocardiography-based STEMI anatomical location was not uniformly documented in electronic medical records during the enrollment period; therefore, we adopted culprit vessel type and multivessel disease as objective alternative indicators to reflect infarct-related ischemic characteristics, which were fully adjusted in multivariate models. In addition, dynamic myocardial enzyme measurements were not incorporated due to inconsistent admission and blood-sampling time windows among emergency STEMI patients with heterogeneous disease onset times. Instead, uniformly recorded peak cTnI was used to quantitatively reflect the maximum severity of myocardial necrosis, which is the most reliable biomarker for evaluating myocardial injury and predicting post-procedural heart failure. Future prospective studies with standardized unified sampling time points and complete infarct location recording are needed to further validate our findings. Second, admission glucose represented a single measurement (at the emergency department time point) and may not fully represent glycemic status throughout hospitalization. Continuous glucose monitoring (CGM) or multi-timepoint glucose measurements might provide more comprehensive glucose exposure information. Third, HbA1c as the basis for eAG derivation has limitations: in patients with concurrent anemia, hemoglobinopathies, recent blood transfusions, or severely impaired hepatic or renal function, HbA1c may not accurately reflect chronic glycemic levels [40]. Although this study excluded extreme HbA1c values (< 4.0% and > 14.0%) and patients with severe hepatic dysfunction, these confounders could not be completely eliminated. Fourth, the definition of in-hospital HF, while incorporating multiple dimensions (Killip class, clinical diagnosis, imaging, and hemodynamics), inevitably involves information bias due to its retrospective determination. Prospective studies should employ standardized HF diagnostic protocols and core laboratory adjudication. Fifth, the lack of external validation cohorts represents a limitation. The optimal cutoff value (SHR=1.27) and predictive model derived from this study require external validation in independent multicenter cohorts before widespread application. Sixth, restricted cubic spline analysis revealed markedly widened 95% confidence bands at the extreme upper tail of the SHR distribution (SHR > 2.5), which is attributed to obvious data sparsity and limited sample size among patients with extremely high SHR values. Such sparse data distribution increases statistical sampling uncertainty and reduces the quantitative predictive accuracy of the model for extreme high-risk outliers. Nevertheless, the confidence intervals remained narrow and stable across the major SHR distribution range (0.5–2.5), strongly supporting the robust linear dose-response relationship between SHR and in-hospital HF risk in the vast majority of STEMI patients; this limitation only affects extreme outliers and does not compromise the core conclusion of the present study. In addition, since SHR is a secondary derived index based on admission glucose and HbA1c, its ROC predictive performance has certain interpretative limitations and cannot provide completely independent biomarker information beyond the original glycemic indicators. The superior AUC of SHR relative to admission glucose only reflects the prognostic improvement effect of chronic glycemic calibration, rather than the discovery of a new predictive biomarker. Future studies are warranted to explore novel independent stress biomarkers to further refine risk stratification for post-STEMI HF.
In conclusion, the present study demonstrates that SHR is a robust, independent predictive biomarker for in-hospital heart failure in STEMI patients undergoing emergency PCI. By eliminating the confounding interference of individual chronic glycemic status, SHR exhibits superior predictive accuracy and incremental prognostic value compared with conventional admission glucose. The linearly increased HF risk alongside elevated SHR, as well as the consistent predictive efficacy across diverse clinical subgroups, further validates its reliable clinical application value. With a favorable discriminatory ability and clear risk stratification cutoff, SHR can serve as a convenient and effective tool for early in-hospital risk assessment. Routine calculation of SHR at admission may help clinicians identify high-risk STEMI patients who are vulnerable to post-PCI heart failure, formulate individualized intensive monitoring and intervention strategies, and ultimately improve the in-hospital clinical prognosis of STEMI patients.
Funding Statement
The author(s) declared that financial support was not received for this work and/or its publication.
Footnotes
Edited by: Nikos Stalikas, OLV Aalst, Belgium
Reviewed by: Hongwei Li, Capital Medical University, China
Hamidreza Soleimani, Bristol Heart Institute, United Kingdom
Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.
Ethics statement
The studies involving humans were approved by the Ethics Committee of Anhui Provincial Hospital. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and institutional requirements.
Author contributions
FS: Formal analysis, Writing – review & editing. WD: Conceptualization, Writing – original draft, Writing – review & editing. CuF: Data curation, Writing – review & editing. CaF: Conceptualization, Writing – review & editing.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
- 1.Ibánez B, James S, Agewall S, Antunes MJ, Bucciarelli-Ducci C, Bueno H. 2017 ESC guidelines for the management of acute myocardial infarction in patients presenting with ST-segment elevation. Rev Esp Cardiol Engl Ed. (2017) 70(12):1082–5. 10.1016/j.rec.2017.11.010 [DOI] [PubMed] [Google Scholar]
- 2.Tsao CW, Aday AW, Almarzooq ZI, Anderson CAM, Arora P, Avery CL, et al. 2023 Update: a report from the American Heart Association. Circulation. (2023) 147(8):e93–621. 10.1161/CIR.0000000000001123 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Bahit MC, Lopes RD, Clare RM, Newby LK, Pieper KS, Van de Werf F, et al. Heart failure complicating non-ST-segment elevation acute coronary syndrome: timing, predictors, and clinical outcomes. JACC Heart Fail. (2013) 1(3):223–9. 10.1016/j.jchf.2013.02.007 [DOI] [PubMed] [Google Scholar]
- 4.Desta L, Jernberg T, Löfman I, Hofman-Bang C, Hagerman I, Spaak J, et al. Incidence, temporal trends, and prognostic impact of heart failure complicating acute myocardial infarction. The SWEDEHEART registry (Swedish web-system for enhancement and development of evidence-based care in heart disease evaluated according to recommended therapies): a study of 199,851 patients admitted with index acute myocardial infarctions, 1996 to 2008. JACC Heart Fail. (2015) 3(3):234–42. 10.1016/j.jchf.2014.10.007 [DOI] [PubMed] [Google Scholar]
- 5.Steg PG, Dabbous OH, Feldman LJ, Cohen-Solal A, Aumont M-C, López-Sendón J, et al. Determinants and prognostic impact of heart failure complicating acute coronary syndromes: observations from the global registry of acute coronary events (GRACE). Circulation. (2004) 109(4):494–9. 10.1161/01.CIR.0000109691.16944.DA [DOI] [PubMed] [Google Scholar]
- 6.de Carvalho LP, Gao F, Chen Q, Sim L-L, Koh T-H, Foo D, et al. Long-term prognosis and risk heterogeneity of heart failure complicating acute myocardial infarction. Am J Cardiol. (2015) 115(7):872–8. 10.1016/j.amjcard.2015.01.010 [DOI] [PubMed] [Google Scholar]
- 7.Ezekowitz JA, Kaul P, Bakal JA, Armstrong PW, Welsh RC, McAlister FA. Declining in-hospital mortality and increasing heart failure incidence in elderly patients with first myocardial infarction. J Am Coll Cardiol. (2009) 53(1):13–20. 10.1016/j.jacc.2008.08.067 [DOI] [PubMed] [Google Scholar]
- 8.Capes SE, Hunt D, Malmberg K, Gerstein HC. Stress hyperglycaemia and increased risk of death after myocardial infarction in patients with and without diabetes: a systematic overview. Lancet (London, England). (2000) 355(9206):773–8. 10.1016/S0140-6736(99)08415-9 [DOI] [PubMed] [Google Scholar]
- 9.Kosiborod M, Rathore SS, Inzucchi SE, Masoudi FA, Wang Y, Havranek EP, et al. Admission glucose and mortality in elderly patients hospitalized with acute myocardial infarction: implications for patients with and without recognized diabetes. Circulation. (2005) 111(23):3078–86. 10.1161/CIRCULATIONAHA.104.517839 [DOI] [PubMed] [Google Scholar]
- 10.Mifsud S, Schembri EL, Gruppetta M. Stress-induced hyperglycaemia. Br J Hosp Med (Lond). (2018) 79(11):634–9. 10.12968/hmed.2018.79.11.634 [DOI] [PubMed] [Google Scholar]
- 11.Deedwania P, Kosiborod M, Barrett E, Ceriello A, Isley W, Mazzone T, et al. Hyperglycemia and acute coronary syndrome: a scientific statement from the American Heart Association diabetes committee of the council on nutrition, physical activity, and metabolism. Anesthesiology. (2008) 109(1):14–24. 10.1097/ALN.0b013e31817dced3 [DOI] [PubMed] [Google Scholar]
- 12.Roberts GW, Quinn SJ, Valentine N, Alhawassi T, O'Dea H, Stranks SN, et al. Relative hyperglycemia, a marker of critical illness: introducing the stress hyperglycemia ratio. J Clin Endocrinol Metab. (2015) 100(12):4490–7. 10.1210/jc.2015-2660 [DOI] [PubMed] [Google Scholar]
- 13.Nathan DM, Kuenen J, Borg R, Zheng H, Schoenfeld D, Heine RJ. Translating the A1C assay into estimated average glucose values. Diabetes Care. (2008) 31(8):1473–8. 10.2337/dc08-0545 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Yang Y, Kim T-H, Yoon K-H, Chung WS, Ahn Y, Jeong M-H, et al. The stress hyperglycemia ratio, an index of relative hyperglycemia, as a predictor of clinical outcomes after percutaneous coronary intervention. Int J Cardiol. (2017) 241:57–63. 10.1016/j.ijcard.2017.02.065 [DOI] [PubMed] [Google Scholar]
- 15.Xu W, Song Q, Wang X, Zhao Z, Meng X, Xia C, et al. Association of stress hyperglycemia ratio and in-hospital mortality in patients with coronary artery disease: insights from a large cohort study. Cardiovasc Diabetol. (2022) 21(1):217. 10.1186/s12933-022-01645-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Li X, Qiao Y, Ruan L, Xu S, Fan Z, Liu S, et al. Stress hyperglycemia ratio as an independent predictor of acute kidney injury in critically ill patients with acute myocardial infarction: a retrospective U.S. Cohort study. Renal Fail. (2025) 47(1):2471018. 10.1080/0886022X.2025.2471018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Liao W-I, Lin C-S, Lee C-H, Wu Y-C, Chang W-C, Hsu C-W, et al. An elevated glycemic gap is associated with adverse outcomes in diabetic patients with acute myocardial infarction. Sci Rep. (2016) 6:27770. 10.1038/srep27770 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Thygesen K, Alpert JS, Jaffe AS, Chaitman BR, Bax JJ, Morrow DA, et al. Fourth universal definition of myocardial infarction (2018). J Am Coll Cardiol. (2018) 72(18):2231–64. 10.1016/j.jacc.2018.08.1038 [DOI] [PubMed] [Google Scholar]
- 19.Avranas K, Mittag M, Schadow K, Eck K, Lehmann R. Impact of the killip class of heart failure on treatment times and intrahospital mortality among STEMI patients. J Cardiovasc Med (Hagerstown). (2025) 26(5):240–7. 10.2459/JCM.0000000000001719 [DOI] [PubMed] [Google Scholar]
- 20.Ponikowski P, Voors AA, Anker SD, Bueno H, Cleland JGF, Coats AJS, et al. 2016 ESC guidelines for the diagnosis and treatment of acute and chronic heart failure. Rev Esp Cardiol (Engl Ed). (2016) 69(12):1167. 10.1016/j.rec.2016.11.005 [DOI] [PubMed] [Google Scholar]
- 21.van Diepen S, Katz JN, Albert NM, Henry TD, Jacobs AK, Kapur NK, et al. Contemporary management of cardiogenic shock: a scientific statement from the American Heart Association. Circulation. (2017) 136(16):e232–e68. 10.1161/CIR.0000000000000525 [DOI] [PubMed] [Google Scholar]
- 22.Austin PC. An Introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. (2011) 46(3):399–424. 10.1080/00273171.2011.568786 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Pencina MJ, D'Agostino RB, Sr., D'Agostino RB, Jr., Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med. (2008) 27(2):157–72. 10.1002/sim.2929 discussion 207-12. [DOI] [PubMed] [Google Scholar]
- 24.Brownlee M. Biochemistry and molecular cell biology of diabetic complications. Nature. (2001) 414(6865):813–20. 10.1038/414813a [DOI] [PubMed] [Google Scholar]
- 25.Chatham JC, Young ME. Metabolic remodeling in the hypertrophic heart: fuel for thought. Circ Res. (2012) 111(6):666–8. 10.1161/CIRCRESAHA.112.277392 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Hegab Z, Gibbons S, Neyses L, Mamas MA. Role of advanced glycation end products in cardiovascular disease. World J Cardiol. (2012) 4(4):90–102. 10.4330/wjc.v4.i4.90 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Przyklenk K, Maynard M, Greiner DL, Whittaker P. Cardioprotection with postconditioning: loss of efficacy in murine models of type-2 and type-1 diabetes. Antioxid Redox Signaling. (2011) 14(5):781–90. 10.1089/ars.2010.3343 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Son SM, Whalin MK, Harrison DG, Taylor WR, Griendling KK. Oxidative stress and diabetic vascular complications. Curr Diab Rep. (2004) 4(4):247–52. 10.1007/s11892-004-0075-8 [DOI] [PubMed] [Google Scholar]
- 29.Neri M, Cerretani D, Fiaschi AI, Laghi PF, Lazzerini PE, Maffione AB, et al. Correlation between cardiac oxidative stress and myocardial pathology due to acute and chronic norepinephrine administration in rats. J Cell Mol Med. (2007) 11(1):156–70. 10.1111/j.1582-4934.2007.00009.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Aljada A, Ghanim H, Mohanty P, Syed T, Bandyopadhyay A, Dandona P. Glucose intake induces an increase in activator protein 1 and early growth response 1 binding activities, in the expression of tissue factor and matrix metalloproteinase in mononuclear cells, and in plasma tissue factor and matrix metalloproteinase concentrations. Am J Clin Nutr. (2004) 80(1):51–7. 10.1093/ajcn/80.1.51 [DOI] [PubMed] [Google Scholar]
- 31.Vandanmagsar B, Youm Y-H, Ravussin A, Galgani JE, Stadler K, Mynatt RL, et al. The NLRP3 inflammasome instigates obesity-induced inflammation and insulin resistance. Nat Med. (2011) 17(2):179–88. 10.1038/nm.2279 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Vaidyula VR, Boden G, Rao AK. Platelet and monocyte activation by hyperglycemia and hyperinsulinemia in healthy subjects. Platelets. (2006) 17(8):577–85. 10.1080/09537100600760814 [DOI] [PubMed] [Google Scholar]
- 33.Rao AK, Chouhan V, Chen X, Sun L, Boden G. Activation of the tissue factor pathway of blood coagulation during prolonged hyperglycemia in young healthy men. Diabetes. (1999) 48(5):1156–61. 10.2337/diabetes.48.5.1156 [DOI] [PubMed] [Google Scholar]
- 34.Pandolfi A, Giaccari A, Cilli C, Alberta MM, Morviducci L, De Filippis EA, et al. Acute hyperglycemia and acute hyperinsulinemia decrease plasma fibrinolytic activity and increase plasminogen activator inhibitor type 1 in the rat. Acta Diabetol. (2001) 38(2):71–6. 10.1007/s005920170016 [DOI] [PubMed] [Google Scholar]
- 35.Lopaschuk GD, Ussher JR, Folmes CD, Jaswal JS, Stanley WC. Myocardial fatty acid metabolism in health and disease. Physiol Rev. (2010) 90(1):207–58. 10.1152/physrev.00015.2009 [DOI] [PubMed] [Google Scholar]
- 36.Taegtmeyer H, McNulty P, Young ME. Adaptation and maladaptation of the heart in diabetes: part I: general concepts. Circulation. (2002) 105(14):1727–33. 10.1161/01.CIR.0000012466.50373.E8 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.