Blog

Introduction

Sepsis, a life-threatening organ dysfunction syndrome had affected over 19 million individuals globally each year, with a case fatality rate exceeding 25%.^1,2 Notably, current epidemiological data predominantly derive from healthcare databases in high-income countries, while low- and middle-income countries (eg, China) face substantial systematic data gaps.³ This study innovatively develops a universal dynamic assessment tool to establish an objective quantification method for the systematically underestimated sepsis burden. Concurrently, it aims to provide actionable decision support for primary medical institutions, with the ultimate goal of reducing avoidable sepsis-related deaths.⁴

The Sequential Organ Failure Assessment (SOFA) score, recognized as the gold standard for evaluating sepsis severity, exhibits a positive correlation with mortality—with higher scores indicating an increased risk of death.⁵ This scoring system is primarily utilized in intensive care unit (ICU) settings to evaluate and monitor critically ill patients with suspected or confirmed sepsis, multisystem organ dysfunction, or septic shock.⁶ In non-ICUs (eg, Infectious Disease wards), the absence of ICU-grade monitoring systems for continuous hemodynamic profiling makes it challenging to establish effective sepsis diagnostic tools.⁷ Furthermore, traditional scoring systems have not integrated dynamically changing biomarkers such as heparin-binding protein (HBP) and interleukin-6 (IL-6), leading to delays in diagnosis and increased risk of false negatives.⁸ Given these persistent monitoring gaps, efforts must be focused on biomarker combinations that circumvent the conventional physiological parameter dependencies.

In biomarker research, combined analysis of IL-6, IL-8, and HBP enhanced early sepsis detection and severity evaluation, complementing traditional markers, such as C-reactive protein(CRP) and procalcitonin(PCT).^9,10 IL-6 also were reported which include inhibition of the release of TNF-βand IL-1 and increasing the levels of anti-inflammatory mediators in circulation which considered as important biomarkers of immune homeostasis.¹¹ The combined model (HBP+PCT+ CRP+IL-6+SOFA) boosted sepsis diagnosis accuracy compared to SOFA alone, which has been validated in multicenter studies.^12–14 D-dimer serves as a key prognostic biomarker for sepsis, predicting 28-day mortality and treatment response via linkage to coagulation disorders and microthrombosis.¹⁵ It should be noted that comorbidity spectrum (such as chronic liver disease, chronic kidney disease) directly affects immune response and coagulation function, exacerbating organ damage in sepsis and neglecting such factors can introduce bias into risk stratification.¹⁶ Current models focus on static data and lack the AI-driven integration of real-time vital signs, biomarkers, and omics data, hindering precise sepsis diagnosis.¹⁷

Notably, existing drug targets (Protein C and Thrombopoietin) exhibit limited efficacy, underscoring the need to identify novel pathways (including endothelial injury and mitochondrial dysfunction) via innovative combinations of biomarkers to facilitate the development of broader-spectrum therapeutics.¹⁸

This study presents an effective framework that systematically integrates real-time kinetic laboratory indices with comprehensive comorbidity profiling, using a dynamic nomogram. Continuous kinetics profiling in sepsis overcomes static data constraints to guide dynamic therapeutic stratification.¹⁹ By engineering this diagnostic instrument, we aimed to revolutionize sepsis diagnostic paradigms using precision-based risk stratification, particularly in non-intensive care unit settings.

Materials and Methods

Study Design and Participants

We collected cases from all inpatients of the Infectious Diseases Departments of two hospitals affiliated with Taizhou Enze Medical Center as the research subjects from January 2024 to December 2024. Patients were excluded if they had: 1) incomplete clinical data precluding accurate assessment of infection or organ dysfunction, 2) missing values for any critical laboratory indicators.

A total of 1,098 patients admitted to Taizhou Hospital of Zhejiang Provincial from January 2024 to December 2024 were finally enrolled in the study. Additionally, 94 patients from Enze Hospital were recruited as the validation cohort between January 2024 and March 2024. G*Power analysis demonstrated statistical powers of 0.99 in the model cohort and 0.85 in the validation cohort, satisfying the predefined power threshold requirements. Sepsis was defined according to the the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3).²⁰ The diagnostic criteria for sepsis required: 1) confirmed or suspected invasive infection and 2) fulfillment of ≥2 systemic inflammatory response syndrome (SIRS) parameters (abnormalities in core temperature, leukocyte count, heart rate, and respiratory rate). All final diagnoses were independently verified by two board-certified infectious disease specialists (Cohen’s Kappa κ=0.85).

Sample and Laboratory Analysis

Peripheral venous blood samples were collected from all cohorts under fasting conditions within 24h of admission. Blood counts were measured using a Mindray BC series automatic blood cell analyzer (Mindray, Shenzhen, China). Routine blood coagulation tests were performed using an STR-Max automatic coagulation analyzer (Stago, Cedex, France). Biochemical indicators were detected using an AU5800 Beckman Library automatic biochemical analyzer (Beckman Coulter, Brea, CA, USA). PCT and IL-6 levels were quantitatively determined using a Cobas 8000 electrochemiluminescence analyzer (Roche Diagnostics, Basel, Switzerland). HBP levels were quantitatively analyzed using a Jet-iStar 3000 dry fluorescence immunoassay system (Zhonghan Shengtai Biotechnology Co., Ltd., Shenzhen, China). All assays were performed using manufacturer-provided reagents in accordance with the standardized operational guidelines.

Clinical Data Collection

Longitudinal clinical metrics, including demographics, comorbid conditions, infectious etiologies, organ dysfunction, and length of hospital stay (LOS), were systematically collected from medical records. To reduce information bias, this study used blinded data extraction: two researchers independently extracted key variables (eg, age, laboratory indicators) from medical records, cross-checked data, and resolved inconsistencies through medical record committee discussion. During statistical analysis, group information was coded (eg, sepsis Group and non-sepsis Group), with analysts blinded to actual group assignments until the completion.

Statistical Analysis

Prior to data collection, statistical power calculation was conducted using G*Power 3.1 (Heinrich Heine University, Düsseldorf, Germany) to ensure sufficient statistical sensitivity, with the β-level constrained to ≥0.70 threshold. All computational procedures were implemented using the R software (v4.4.2, R Foundation). Non-normally distributed continuous measures were summarized as medians with interquartile ranges and analyzed using the Wilcoxon rank-sum test. Categorical parameters are presented as counts with proportional distributions, and between-group differences were assessed using Pearson’s χ²-test.

Multivariable regression modeling incorporating variables identified through preliminary analyses enabled model construction. The diagnostic performance of the models was evaluated using ROC curve analysis with the corresponding Area Under the Curve (AUC) quantification. Subsequently, a clinical diagnosis tool (web-based dynamic nomogram) was developed based on the regression coefficients. The predictive accuracy was assessed through concordance index calculations, and calibration curves were used to evaluate the model fit. The clinical utility was further examined using decision curve analysis. The statistical significance threshold was set at p<0.05.

Ethics Approval and Informed Consent

This study is a retrospective analysis of patients’ historical laboratory data. Prior to hospitalization, all patients were informed and provided written consent to the following:

To advance medical research and education, your medical records along with residual biological specimens (including blood, bodily fluids, tissues, etc.) may be utilized. We undertake to maintain strict confidentiality of your personal information throughout this process.

Enze Medical Center Group has an ethics committee named after Taizhou Hospital of Zhejiang Province. This study was conducted in accordance with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of Taizhou Hospital of Zhejiang Province (KL20240640). This ethics committee applies to all its affiliated hospitals, among which Enze Hospital is a branch.

Results

Clinical Characteristics of the Patients

Research subjects were divided into sepsis (n=354) and non-sepsis (n=744) groups (Figure 1). The median age of the study population was 64 years (IQR 51–73) with a male predominance (56.74%). The most common comorbidity was hypothyroidism (60.66%), followed by renal insufficiency (44.63%), hypertension (39.34%), diabetes (20.77%), malignant tumors (20.31%), brain dysfunction (16.94%, and cardiac insufficiency (14.48%). Comparative analysis revealed significant age disparity between groups, with sepsis patients exhibiting an older age distribution (median 67 years, IQR 57–75) compared to non-sepsis patients (median 62 years, IQR 50–72; p<0.001). Notably, patients with sepsis were more likely to have renal disease (57.34% vs 38.58%, p<0.001), hypertension (51.41% vs 33.60%, p<0.001), and cardiac insufficiency (21.47% vs 11.16%, p<0.001). The sepsis group demonstrated a significantly extended LOS than the non-sepsis group [6(5, 9) vs 6(4, 9), p=0.017] (Table 1).

Table 1 Baseline Characteristics of the Patients

Figure 1 Study flowchart.

Laboratory Data of the Patients

Comparative evaluation of the baseline hematological parameters between the groups revealed distinct inflammatory profiles (Table 1). The sepsis group demonstrated significant elevations in established inflammatory biomarkers: CRP (33.70, 8.10–101.00 vs 23.20, 6.00–74.40; p=0.002), and PCT (0.77, 0.23–4.69 vs 0.22, 0.08–0.69; p<0.001), HBP (229.49, 113.56–300.00 vs 179.57, 83.65–300.00; p<0.001). IL-6 concentrations showed marked intergroup differences (median 84.15, IQR 77.40–541.00 vs 77.40, 44.60–77.40; p<0.001).

Cellular immune profiling indicated neutrophilic predominance in the sepsis group, as evidenced by an elevated neutrophil percentage (77.80%, 64.40–88.50 vs 71.80%, 60.70–82.85; p<0.001), NLR (6.01, 2.77–14.29 vs 3.85, 2.26–7.97; p<0.001), PDW (16.30,16.00–16.70 vs 16.20, 15.90–16.50); p<0.001), MPV (9.60,9.00–10.50 vs 9.45, 8.80–10.30; p<0.001) and PLR (155.79, 100.77–270.40 vs 145.91, 95.89–217.33); p=0.046).

Compared with the non-sepsis group, the sepsis group exhibited significantly suppressed lymphocyte and monocyte percentages, with median values decreasing from 18.30% (IQR: 10.35–27.40) to 12.90% (6.30–23.40), p< 0.001) and from 6.70% (5.00–8.60) to 5.85% (4.00–8.00), p< 0.001), respectively. Notably, even more marked reductions were observed in eosinophil (0.55% [0.10–2.00] vs 0.90% [0.20–2.20], p< 0.001) and basophil percentage (0.10% [0.10–0.30] vs 0.20% [0.10–0.40], p< 0.001). Additionally, platelet counts were found to be suppressed (157.00 [106.00–220.00] vs 187.00 [127.50–248.50], p< 0.001). Coagulation abnormalities were reflected in elevated D-dimer levels (1.59 [0.72–2.98]vs 0.98 [0.49–1.97] mg/L, p<0.001) in the sepsis group.

Multivariable Logistic Regression for Sepsis Diagnosis

Multivariate logistic regression analysis identified eight independent predictors of sepsis development (Table 2). Clinical comorbidities: Hypertension: OR=1.6278 (95% CI: 1.2079–2.1937, p=0.001), Renal dysfunction: OR=1.7002 (95% CI: 1.2840–2.2513, p=0.002), Cardiac dysfunction: OR=1.8927 (95% CI:1.2979–2.7599, p=0.001); Inflammatory and coagulation biomarkers: IL-6: OR=1.0003 (95% CI: 1.0002–1.0005, p<0.001), Basophil percentage: OR=0.4319 (95% CI: 0.2353–0.7926, p=0.007), D-dimer: OR=1.0796 (95% CI: 1.0273–1.1347, p=0.0025), PLR: OR=1.0025, 95% CI, 1.0011–1.0040), PLT: OR=0.9939, 95% CI, 0.99119–0.995. The final model demonstrated an acceptable discriminative ability [AUC=0.746, 95% CI: (0.709–0.772)]. ROC analysis further revealed a negative predictive value (NPV) of 0.832 and a positive predictive value (PPV) of 0.511.

Table 2 Univariate and Multivariate Analysis for Diagnosis of Sepsis

Establishment and Accuracy Diagnosis of a Nomogram

We developed a nomogram that integrates eight elements for the diagnosis of sepsis (Figure 2A). Through internal validation, the calibration curve (Figure 2B) showed no significant deviation from the reference line; the decision curve (Figure 2C) also demonstrated the nomogram had favorable net benefits for disease identification. The Web-based interactive nomogram for sepsis management is now available on a secure platform.

Figure 2 Establishment and accuracy diagnosis of nomogram. (A) Nomogram (B) calibration curve (C) Decision curve (D): ROC of the modeling and validation groups.

External Validation

A total of 94 newly enrolled patients were included in the external validation, comprising a sepsis group (n=34) and a non-sepsis group (n=60). The external validation yielded an AUC of 0.663 (95% CI: 0.549, 0.776), with no statistically significant difference in predictive performance observed between the external and internal validation cohort (Figure 2D). This tool can be accessed at: https://bixiaojie-1987.shinyapps.io/DynNomapp/ (Figure 3).

Figure 3 Web-based dynamic nomogram.

Discussion

This study presents an effective web-based dynamic nomogram that enhances sepsis diagnosis in non-ICU settings by integrating clinical comorbidities (including Hypertension, Renal dysfunction, and Cardiac dysfunction) and routine laboratory indicators (such as D-dimer, Basophil percentage, PLT, PLR, and IL-6). Our findings contribute to advancing precision sepsis diagnosis in resource-limited environments where continuous hemodynamic monitoring is infeasible.

Our data identified hypertension, renal and cardiac dysfunction, D-dimer, Basophil percentage, PLT, PLR, and IL-6 was independent risk factors for the development of sepsis in patients with infectious diseases. Patients with chronic hypertension exhibit compromised host defense mechanisms, secondary to persistent vascular endothelial lesions and microcirculatory hemodynamic disturbances.²¹ Renal insufficiency leads to the accumulation of metabolic waste products and electrolyte imbalances, which further exacerbate systemic inflammatory responses and multi-organ dysfunction.²² Cardiac dysfunction results in decreased cardiac output, exacerbating tissue hypoperfusion and microcirculatory impairment, which accelerates the development of multiple organ dysfunction syndrome.²³ Clinicians should maintain heightened vigilance for early recognition of sepsis in patients with chronic hypertension, renal insufficiency, or cardiac dysfunction.dimer concentrations were substantially elevated in sepsis patients compared to non-septic controls. D-dimer, a degradation product generated during the coagulation cascade, is a critical biomarker for predicting adverse prognostic outcomes in sepsis patients.²⁴ Patients with severe sepsis frequently succumb to diffuse intravascular coagulation (DIC), which often progresses to multi-organ failure through thrombotic microangiopathy and consumption coagulopathy.²⁵ Immune profiling revealed a significant reduction in basophil percentage among sepsis patients (OR = 0.4319), consistent with established evidence of basophil depletion as a prognostic indicator in sepsis. Basophils mitigate the cytokine storm in sepsis by releasing cytokines, such as IL-4 and IL-13, which inhibit excessive inflammatory responses. This observation may be explained by inflammatory cytokine-mediated suppression of bone marrow egress. Basophil depletion has emerged as a robust independent prognostic factor in sepsis patients.²⁶ Elevated D-dimer levels, along with Basophil, can serve as promising biomarkers to guide clinical decision-making and facilitate the formulation of personalized therapeutic strategies in patients with sepsis.

Thrombocytopenia on ICU admission is a well-recognized marker of poor prognosis in septic patients.²⁷ PLT reduction is often the “first signal” of coagulation dysfunction in sepsis, detectable even when the SOFA score has not yet worsened. IL-6 significantly increases 2–4 hours after infection, earlier than CRP or PCT. A high PLR usually indicates more severe conditions and poorer prognosis in sepsis patients, and can serve as a simple auxiliary indicator for evaluating sepsis in clinical practice.²⁸ PLR can sensitively capture the immune-inflammatory status of sepsis through the “dual imbalance” of platelet activation (pro-inflammatory) and lymphocyte depletion (immunosuppression). In summary, even in the early stages of hospitalization, the high sensitivity indicators can quickly “capture” potential severe patients, even if the OR values of the three are low.

The web-based dynamic nomogram developed in this study overcomes the limitations of traditional predictive tools owing to its networked architecture.^29,30 Unlike the SOFA score, which relies on ICU-specific monitoring parameters (eg, lactate levels, vasoactive medication use) and requires manual entry of complex data, our model offers distinct practical advantages: it integrates laboratory indicators with additional inflammatory markers and features a dynamic web-based interface enabling real-time calculations on mobile devices.

Notably, the model demonstrates robust predictive performance—particularly its strong negative predictive value—making it suitable for initial sepsis risk assessment in general ward settings. This application could help mitigate unnecessary ICU admissions while prompting intensified monitoring (eg, lactate measurement) for high-risk patients. Furthermore, to facilitate broader utility and future optimization, we have made the model code publicly available via networked deployment. With appropriate access permissions, researchers can enhance the model by incorporating novel laboratory indicators or refining existing parameters.

This study had some limitations that should be considered. First, the model was trained and validated primarily on retrospective data from a single tertiary care center, which may introduce selection bias due to overrepresentation of severe sepsis cases. Second, the model has not yet been integrated into clinical application platforms; as such, its practical utility in real-world clinical workflows remains to be verified. For future research, efforts should focus on multicenter validation to enhance generalizability, and more external laboratory indicators should be collected to identify additional biomarkers. Finally, the model will be deployed in the official hospital app to facilitate its translation into clinical practice and assess its performance in real-time clinical decision-making scenarios.

Conclusion

This study introduces an effective dynamic nomogram that improves sepsis diagnosis in non-ICU environments by integrating clinical comorbidities and routine laboratory indicators. This tool addresses systemic limitations in resource-constrained settings and offers a scalable solution for early detection and risk stratification of sepsis. The integration of readily available clinical data ensures practical applicability without requiring specialized equipment, thereby facilitating timely interventions and potentially reducing sepsis-related mortality. The model should undergo further optimization and validation through networking to fully benefit clinical practice.

Acknowledgments

We would like to thank the nurses in Infectious Diseases Departments of Taizhou Enze Medical Center for sampling the specimens and the patients for enrollment in this study.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Funding

Medical Science and Technology Project of Zhejiang Province 2024KY522.

Disclosure

The authors declare no competing interests in this work.

References

1. Rudd KE, Johnson SC, Agesa KM, et al. Global, regional, and national sepsis incidence and mortality, 1990-2017: analysis for the global burden of disease study. Lancet. 2020;395(10219):200–211. doi:10.1016/S0140-6736(19)32989-7

2. Giamarellos-Bourboulis EJ, Aschenbrenner AC, Bauer M, et al. The pathophysiology of sepsis and precision-medicine-based immunotherapy. Nat Immunol. 2024;25(1):19–28. doi:10.1038/s41590-02

3. Minderhoud TC, Azijli K, Nanayakkara PWB. Sepsis. Lancet. 2020;396(10265):1804. doi:10.1016/S0140-6736(20)32401-6

4. Lorencio Cárdenas C, Yébenes JC, Vela E, et al. Trends in mortality in septic patients according to the different organ failure during 15 years. Crit Care. 2022;26(1):302. doi:10.1186/s13054-022-041

5. Bonini A, Carota AG, Poma N, et al. Emerging biosensing technologies towards early sepsis diagnosis and management. Biosensors. 2022;12(10):894. doi:10.3390/bios12100894

6. Bouwman W, Verhaegh W, Stolpe AVD. androgen receptor pathway activity assay for sepsis diagnosis and prediction of favorable prognosis. Front Med-Lausanne. 2021;8:767145. doi:10.3389/fmed.2021.767145

7. Wang X, Guo Z, Chai Y, et al. Application Prospect of the SOFA Score and Related Modification Research Progress in Sepsis. J Clin Med. 2023;12(10):3493. doi:10.3390/jcm12103493

8. Blanchard F, Charbit J, Van der Meersch G, et al. Early sepsis markers in patients admitted to intensive care unit with moderate-to-severe diabetic ketoacidosis. Ann Intensive Care. 2020;10(1):58. doi:10.1186/s13613-020-00676-6

9. Zhang Z, Zhu Y, Cao Y, et al. Predictive value of heparin binding protein for sepsis. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 2021;33(6):654–658. doi:10.3760/cma.j.cn121430-20210424-00605

10. Feng L, Liu S, Wang J, et al. The performance of a combination of heparin-binding protein with other biomarkers for sepsis diagnosis: an observational cohort study. Bmc Infect Dis. 2024;24(1):755. doi:10.1186/s12879-024-09666-6

11. Steensberg A, Cp F, Keller C, Moller K, Pedersen BK. IL-6 enhances plasma IL-1ra, IL-10, and cortisol in humans. Am J Physiol-Endoc M. 2003;285(2):E433–7. doi:10.1152/ajpendo.00074.20

12. Hong X, Wang Y, Ma S, Xu M, Xu Z. Diagnostic value of plasma heparin-binding protein and the heparin-binding protein-to-albumin ratio in patients with community-acquired Pneumonia: a retrospective study. Bmc Infect Dis. 2023;23(1):777. doi:10.1186/s12879-023-08762-3

13. Liu Z, Meng Z, Li Y, et al. Prognostic accuracy of the serum lactate level, the SOFA score and the qSOFA score for mortality among adults with Sepsis. Scandinavian J Trauma Resuscitation Emerg Med. 2019;27(1):51. doi:10.1186/s13049-019-0609-3

14. Li Z, Huang B, Yi W, et al. Identification of potential early diagnostic biomarkers of sepsis. J Inflamm Res. 2021;14:621–631. doi:10.2147/JIR.S298604

15. Tang J, Yuan H, Yl W, Fu S, Pan XY. the predictive value of heparin-binding protein and d-dimer in patients with sepsis. Int J Gen Med. 2023;16(16):2295–2303. doi:10.2147/IJGM.S409328

16. Mira JC, Gentile LF, Mathias BJ, et al. Sepsis pathophysiology, chronic critical illness, and persistent inflammation-immunosuppression and catabolism syndrome. Crit Care Med. 2017;45(2):253–262. doi:10.1097/CCM.0000000000002074

17. Lee J, Song J. Performance of a quick sofa-65 score as a rapid sepsis screening tool during initial emergency department assessment: a propensity score matching study. J Crit Care. 2020;55:1–8. doi:10.1016/j.jcrc.2019.09.019

18. Seymour CW, Gesten F, Prescott HC, et al. Time to treatment and mortality during mandated emergency care for sepsis. New Engl J Med. 2017;376(23):2235–2244. doi:10.1056/NEJMoa17030

19. Lu J, Zhang W, He Y, et al. Multi-omics decodes host-specific and environmental microbiome interactions in sepsis. Front Microbiol. 2025;16:1618177. doi:10.3389/fmicb.2025.1618177

20. Singer M, Deutschman CS, Seymour CW, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA-J Am Med Assoc. 2016;315(8):801. doi:10.1001/jama.201

21. Xu J, Zhang W, Fu J, et al. Viral sepsis: diagnosis, clinical features, pathogenesis, and clinical considerations. Military Med Res. 2024;11(1):78. doi:10.1186/s40779-024-00581-0

22. Kalantar KL, Neyton L, Abdelghany M, et al. Integrated host-microbe plasma metagenomics for sepsis diagnosis in a prospective cohort of critically ill adults. Nat Microbiol. 2022;7(11):1805–1816. doi:10.1038/s41564-022-01237-2

23. Doualeh M, Payne M, Litton E, Raby E, Currie A. Molecular methodologies for improved polymicrobial sepsis diagnosis. Int J Mol Sci. 2022;23(9):4484. doi:10.3390/ijms23094484

24. Lu J, Fang W, Lei Y, Yang J. Association between D-dimer-to-albumin ratio and 28-days all-cause mortality in patients with sepsis. Sci Rep-Uk. 2024;14(1):28361. doi:10.1038/s41598-024-79911-0

25. Grondman I, Pirvu A, Riza A, Ioana M, Netea MG. Biomarkers of inflammation and the etiology of sepsis. Biochem Soc Trans. 2020;48(1):1–14. doi:10.1042/BST20190029

26. Chen X, Zhu X, Zhuo H, Lin J, Lin X. Basophils absence predicts poor prognosis and indicates immunosuppression of patients in intensive care units. Sci Rep-Uk. 2023;13(1):18533. doi:10.1038/s41598-023-45865-y

27. Fan SH, Pang MM, Si M, et al. Quantitative changes in platelet count in response to different pathogens: an analysis of patients with sepsis in both retrospective and prospective cohorts. Ann Med. 2024;56(1):2405073. doi:10.1080/07853890.2024.2405073

28. Wang G, Mivefroshan A, Yaghoobpoor S, et al. prognostic value of platelet to lymphocyte ratio in sepsis: a systematic review and meta-analysis. Biomed Res Int. 2022;2022:9056363. doi:10.1155/2022/9056363

29. Yang Y, Dong J, Li Y, et al. Development and validation of a nomogram for predicting the prognosis in cancer patients with sepsis. Cancer Med-Us. 2022;11(12):2345–2355. doi:10.1002/cam4.4618

30. He Y, Xu J, Shang X, et al. Clinical characteristics and risk factors associated with ICU-acquired infections in sepsis: a retrospective cohort study. Front Cell Infect Mi. 2022;12:962470. doi:10.3389/fcimb.2022.962470

Introduction

Type 2 diabetes mellitus (T2DM) affects 537 million adults globally, with projections exceeding 783 million by 2045.¹ It has attracted much attention in recent years due to its dominant global prevalence (accounting for 90–95% of diabetes cases) and its strong epidemiological link to coronary artery disease (CAD).² Cardiovascular disease (CVD) accounts for 50% of mortality in T2DM, and CAD represents its most lethal manifestation.³ Unlike type 1 diabetes, T2DM involves insulin resistance, chronic inflammation and dyslipidaemia, which synergistically accelerate atherosclerosis.⁴

Prolonged hyperglycaemia drives CAD pathogenesis through multiple pathways: (1) oxidative stress-induced endothelial dysfunction, (2) advanced glycation end-product (AGE)-mediated vascular inflammation and (3) dyslipidaemia-enhanced plaque instability.^5–7 Although intensive glucose control reduced microvascular complications in landmark trials (such as the UK Prospective Diabetes Study), its macrovascular benefits remain contested. The Action to Control Cardiovascular Risk in Diabetes trial reported increased mortality with glycated haemoglobin (HbA1c) <6.5%,⁸ whereas the Action in Diabetes and Vascular Disease: PreterAx and DiamicroN Controlled Evaluation and the Veterans Affairs Diabetes Trial found neutral effects on major CVD events.⁹ This paradox underscores the need for personalised glycaemic targets tailored to factors such as diabetes duration, comorbidities and vascular pathology.^10,11 Given these unique mechanisms and the rising burden of T2DM in ageing populations, clarifying optimal glycaemic targets for this subgroup remains clinically urgent.

Current guidelines lack granularity for patients with T2DM with established CAD. The 2023 American Diabetes Association (ADA) standards advocate HbA1c <7–8% for high-risk patients but omit stratification by coronary stenosis severity.¹² Similarly, the European Society of Cardiology chronic coronary syndrome (CCS) guidelines prioritise lipid and blood pressure control without specifying glucose targets for advanced atherosclerosis.¹³

To bridge this gap, we investigate how glycaemic control, diabetes duration and coronary stenosis severity (quantified by the Gensini score) jointly influence long-term mortality in patients with T2DM–CCS.

Methods

Study Design and Population

This study retrospectively included a total of 390 patients with T2DM and CCS who underwent coronary angiography between 2011 and 2012. The inclusion criteria were as follows: (1) participants aged >18 years; (2) patients clinically diagnosed with T2DM and confirmed to have CAD through coronary angiography. The exclusion criteria were as follows: (1) patients lacking essential laboratory data, such as HbA1c levels; (2) patients with severe liver or kidney dysfunction or malignant tumours; (3) patients with concomitant congenital heart diseases or cardiomyopathies.

After screening, a total of 150 patients were included for final analysis (Figure 1). Patients were stratified into the intensified control group and the relaxed control group based on their HbA1c levels being ≤7.5% or >7.5%, respectively. The HbA1c threshold of 7.5% was selected based on three considerations: (1) the ADA guidelines recommend individualised targets (HbA1c <7.5%) for patients with advanced CVD;¹⁴ (2) this cutoff aligns with studies focusing on high-risk T2DM populations, where stringent control (<6.5%) showed no mortality benefit;¹⁵ (3) it reflects real-world clinical practice balancing hypoglycaemia risk and glycaemic management in comorbid patients.¹⁶

Figure 1 Flow chart.

All reporting adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) standards.¹⁷ All patients were assessed at the Second Affiliated Hospital of Xi’an Medical University through outpatient visits every 6 months or telephone interviews. The clinical information and outcome events were recorded in study-specific databases. The participants were telephoned with their consent and informed of the purpose of the study. In this study, patients or their family members agreed by telephone or verbal consent during outpatient follow-up. They were told that this was a retrospective study and all data were anonymised during the analysis process and did not compromise the privacy of patients; furthermore, the study covered a long period and involved a retrospective analysis of hospitalised patients from up to 10 years ago. Some patients died or were unable to sign informed consent due to complications during the follow-up.

Ethical approval was obtained from the Institutional Review Board of The Second Affiliated Hospital of Xi’an Medical University (No. X2Y202352). The study adhered to the Declaration of Helsinki and STROBE guidelines. All participants or their legal surrogates provided verbal informed consent via telephone interviews. The process of obtaining consent was meticulously documented through audio recordings, which were stored securely and archived. Participants were informed about the purpose of the study, the nature of the data collection, and how their data would be used. It was explicitly stated that all data would be anonymised during the analysis process to ensure the privacy and confidentiality of the participants. Additionally, participants were assured that their participation was voluntary and that they could withdraw from the study at any time without any repercussions.

Variables Collection

The Beckman Coulter Automatic Biochemical Analyzer AU5800 and Beckman Coulter reagents were used to test the baseline data of laboratory measurements, such as low-density lipoprotein cholesterol (LDL-C), triglycerides, total cholesterol, high-density lipoprotein cholesterol (HDL-C), blood glucose, uric acid, total bilirubin (TBIL), direct and indirect bilirubin (DBIL and IBIL, respectively) and serum creatinine. Glycated haemoglobin was measured using the AU5800 analyser with immunoturbidimetric reagents certified by the National Glycohemoglobin Standardization Program. The results of the coronary angiography were interpreted by a cardiologist who had >5 years of working experience. The results of the angiographic CAD were quantified precisely using the Gensini score,¹⁸ which quantifies coronary artery stenosis severity by assigning weights to each lesion based on its location and degree of luminal narrowing; higher scores indicate more extensive coronary disease.

Outcomes

Patients were divided into two categories based on their HbA1c levels at the index date, and the incidence of all-cause and cardiovascular death was compared between the subgroups. The incidence of each component was evaluated in the same way. Any death recorded in the diagnosis procedure combination data for the whole observable period following the index date was considered an all-cause death. Cardiovascular death was defined as myocardial infarction (MI), angina pectoris, malignant arrhythmia, acute heart failure or ischaemic stroke.

Statistical Analysis

All data analyses were performed using the statistical software SPSS 26.0, GraphPad Prism 7 and R version 4.2.2. The median (interquartile, Q1–Q3) was used to express the non-normally distributed data, whereas the mean and standard deviation (SD) were used to portray the variables with a normal distribution. Student’s t-test (data with a normal distribution) and Pearson’s chi-squared test (categorical variables) were run for the comparative analysis between the two groups. The multivariate Cox proportional hazards model was used to compare the risk of endpoints between the HbA1c subgroups. Kaplan–Meier survival analyses with Log rank tests were performed to show differences between groups.

Results

Study Population and Patient Characteristics

Patients were recruited between 1 January 2012 and 31 December 2013, and observation continued until 31 May 2022, for a total of 104.8 ± 35.7 months. In total, 150 patients with T2DM and stable CAD were registered in the cohort. The median diabetic duration was 2.0 years (0.50–10.0 years).

The mean (standard deviation) of age, Hb A1c, follow-up period, fasting glucose, 2-hour postprandial blood glucose, haemoglobin, serum albumin, serum total protein, total cholesterol, creatinine and urea nitrogen at enrolment was 63.5 years (10.8 years), 7.3% (1.8%), 104.8 months (35.7 months), 8.0 mmol/L (2.9 mmol/L), 13.1 mmol/L (4.8 mmol/L), 133 g/L (18.7 g/L), 42.3 g/L (4.3 g/L), 65.6 g/L (5.7 g/L), 1.6 mmol/L (0.8 mmol/L), 81.6 mmol/L (82.5 mmol/L) and 6.7 mmol/L (5.9 mmol/L), respectively. The number (proportion) of men, smokers, hypertension and history of MI or percutaneous coronary intervention were 83 (55.3%), 44 (29.3%), 102 (68%) and 52 (34.7%), respectively.

Hypertension, triglycerides, HDL-C, LDL-C, uric acid, TBIL and DBIL had no statistically significant differences between groups (all p > 0.05). During the observation period, a total of 53 patients died (35%); Table 1 displays the characteristics of the study participants divided into the deceased and surviving groups. Clinical outcomes at 10 years are summarised in Table 2.

Table 1 Characteristics of Study Participants

Table 2 Clinical Outcomes at 10 Years

All-Cause Mortality Risk Factors and Survival Analysis

Multivariate Cox regression analysis showed that all-cause mortality was significantly associated with age, HbA1c, diabetic duration, creatinine and Gensini score (Table 3). Adjusted hazard ratios (HRs) (95% CIs) of all-cause mortality were 1.10 (1.06–1.15) for age, 1.29 (1.12–1.48) for HbA1c, 1.06 (1.02–1.10) for diabetic duration, 1.00 (1.00–1.01) for creatinine and 1.02 (1.01–1.02) for Gensini score. Figure 2 displays the survival curve of different groups according to the level of HbA1c (>7.5% or ≤7.5%). A higher mortality rate was observed in patients with HbA1c of >7.5% among those with diabetic duration >10 years (p = 0.0115, Figure 2a). There was no significant difference between the two groups among those with diabetic duration of <5 years (p = 0.1425, Figure 2b). In the group of Gensini scores of >60 points, no significant differences were detected between patients with HbA1c of >7.5% and ≤7.5% (p = 0.1182, Figure 2c). However, patients with HbA1c of ≤7.5% had less all-cause survival than those with HbA1c of >7.5%, with Gensini scores of ≤60 points (p = 0.0160, Figure 2d). Figure 2d.

Table 3 Cox Survival Analysis of All-Cause Mortality

Figure 2 All-cause mortality risk factors and Survival Analysis. (a) The survival curve of DM duration>10 years group according to the level of HbA1c. (b) The survival curve of DM duration< 5 years group according to the level of HbA1c. (c) The survival curve of Gensini score>60 group according to the level of HbA1c. (d) The survival curve of Gensini score≤60 group according to the level of HbA1c.

Cardiovascular Mortality Risk Factors and Survival Analysis

As shown in Table 4, six variables were selected by the Cox model: age, HbA1c, diabetic duration, DBIL, serum total protein and Gensini score. The adjusted HRs (95% CIs) of cardiovascular mortality were 1.10 (1.05–1.15) for age, 1.36 (1.16–1.58) for HbA1c, 1.06 (1.00–1.10) for the diabetic duration, 1.12 (1.04–1.23) for DBIL, 0.89 (0.83–0.95) for serum total protein and 1.02 (1.01–1.03) for Gensini score. Kaplan–Meier analysis revealed a higher cardiovascular mortality rate in patients with HbA1c of >7.5% and diabetic duration of >10 years (p = 0.0004, Figure 3a). There was no significant difference between the two groups with a diabetic duration of <5 years (p = 0.1340, Figure 3b). Figures 3c and d show that higher mortality rates were observed in patients with HbA1c of >7.5% and Gensini score of >60 points or ≤60 points (p = 0.0154, p = 0.0104).

Table 4 Cox Survival Analysis of Cardiovascular Death

Figure 3 Cardiovascular mortality risk factors and Survival Analysis. (a) The survival curve of DM duration>10 years group according to the level of HbA1c. (b) The survival curve of DM duration< 5 years group according to the level of HbA1c. (c) The survival curve of Gensini score>60 group according to the level of HbA1c. (d) The survival curve of Gensini score≤60 group according to the level of HbA1c.

Discussion

In this retrospective analysis with a 10-year follow-up, in the population with a diabetes course of >10 years, the risk of cardiovascular and all-cause death was substantially lower in the intensive group (HbA1c ≤7.5%) than in the relaxed glycaemic group (HbA1c >7.5%). Our finding that intensive glycaemic control reduced cardiovascular mortality across all stenosis levels (Gensini >60 and ≤60) but improved all-cause survival only in milder stenosis aligns with the ’vascular vulnerability’ hypothesis. In advanced atherosclerosis (Gensini >60), plaques exhibit thin fibrous caps, necrotic cores and macrophage infiltration. This observation indicates that in patients with highly narrowed coronary arteries, solely focusing on blood glucose control may not be sufficient to improve prognosis comprehensively; it may be necessary to consider other risk factors, such as blood pressure and lipid levels, and their management. Here, aggressive HbA1c-lowering may fail to reverse structural instability, and hypoglycaemia episodes could trigger catecholamine surges, plaque rupture and fatal arrhythmias.¹⁹ Conversely, in early-stage CAD (Gensini ≤60), glucose control mitigates endothelial nitric oxide synthase uncoupling, suppresses AGE-RAGE signalling and stabilises plaque phenotype.²⁰ Furthermore, this observation also indicates that in more advanced stages of the disease process, the potential benefit of blood glucose control on overall survival may be somewhat limited, underscoring the need for a more comprehensive treatment approach in clinical practice.

In our study, we also found that the Gensini score was independently associated with mortality in patients with T2DM and CCS. More recently, a retrospective cohort study showed that patients with high Gensini scores had higher event rates for all-cause mortality than those with low Gensini scores.¹⁸ Severe stenosis (Gensini >60) reflects advanced atherosclerosis, where plaque instability and ischaemia-reperfusion injury amplify systemic inflammation. Reynolds et al²¹ have reported that the Gensini score was a highly important predictor of all-cause and cardiac mortality in patients with CAD. Our study results reaffirmed that the Gensini score was an important predictor of a worse prognosis for patients with stable CAD.

Our finding is consistent with previous studies that age is a well-known risk factor for all-cause and CAD mortality; older patients had higher prevalence rates than younger patients.²² Another study has also shown that the rate of mortality significantly increased with diabetes duration.²³ For individuals with T2DM, advancing age often complicates the disease course, as prolonged hyperglycaemia leads to cumulative vascular damage, increasing the risk of CVD. Elderly patients are more prone to developing multiple complications such as hypertension, dyslipidaemia and renal insufficiency, all of which further exacerbate the progression of CVD. Consequently, elderly individuals with T2DM face a relatively higher risk of cardiovascular mortality.

To the best of our knowledge, this is the first study that evaluated the associations of creatinine with total and cardiovascular mortality in the T2DM and CAD population. A retrospective cohort study has shown that a small creatinine increase had a high HR for all-cause mortality (HR 1.577, CI 1.329–1.871) in patients with CVD.²⁴ Previous studies have shown that patients with DM usually have higher serum creatinine and could develop chronic kidney disease.^25,26 The results of this study also demonstrate that creatinine levels are independently associated with all-cause mortality in individuals with T2DM and CAD.

Our study found that in T2DM with CCS, a higher DBIL level, rather than TBIL or IBIL, was positively associated with cardiovascular mortality. Elevated DBIL may indicate impaired hepatic conjugation capacity, linked to oxidative stress and endothelial dysfunction in T2DM. The association between serum bilirubin and T2DM with CAD has been reported in several studies, although the results are inconsistent. Serum bilirubin is a potent endogenous antioxidant and has been identified as a cardiovascular risk in cohort studies,²⁷ although the relation to T2DM and CAD remains unclear. The serum TBIL, which is the sum of the DBIL and IBIL, was the primary focus of previous studies. The meta-analysis reported that TBIL had an inverse association with adverse metabolic outcomes and confirmed a protective role of bilirubin in vascular disease outcomes, such as CAD.²⁸ In contrast, the Dongfeng–Tongji cohort study reported that DBIL concentrations were positively associated with the risk of incident T2DM in middle-aged and elderly adults,²⁹ which was consistent with our results. The controversy in the studies may be due to both genetics and the environment. Additionally, in our study, we found that serum total protein was a protective factor for cardiovascular mortality in T2DM with CCS.

In our study, we also first proposed the glycaemic control strategy using the duration of diabetes and the degree of coronary stenosis. In the ADA guidelines, glycaemic targets are suggested by the number and severity of comorbidities.³⁰ Previous studies have reported that prolonged exposure to hyperglycaemia coupled with several cardiovascular risk factors raises mortality in patients with long-duration diabetes.³¹ We found that long diabetes duration (≥10 years) and strict glycaemic control (≤7.5%) were associated with an decreased risk of death (Figures 2a and 3a). However, for individuals with short diabetes duration (<5 years), good glycaemic control did not further decrease the risk of death. In this study, we also found that severe coronary stenosis (Gensini score >60) and strict glycaemic control were associated with decreased risk of cardiovascular mortality but not all-cause mortality (Figures 2c and 3c), whereas tight glycaemic control further reduced the risk of death among patients with milder coronary stenoses (Gensini score ≤60). According to prior research, targets for glycaemic control should be tailored to each person’s fragility or functional dependency and life expectancy.^32,33 The present study extends prior research by demonstrating that glycaemic control should consider not only the diabetic duration but also the severity of CAD.

This study has several strengths and limitations. It is the first retrospective study that focuses on mortality and glycaemic control based on coronary stenosis and diabetes duration among Chinese people with long-term follow-up (10 years). The following are some of the current study’s flaws: (1) This is a single-centre retrospective study, and the sample size is relatively small. Further large clinical studies are required to validate our findings; (2) the fluctuating relationship over time between HbA1c levels and all-cause mortality was not examined – HbA1c values at various time points would more accurately reflect the risk of mortality because HbA1c fluctuates throughout time. Also, we were unable to obtain complete baseline HbA1c data, glycemic variability or trajectories (for example, patients improving from high to low HbA1c or vice versa) or perform a correlation analysis between baseline HbA1c and the Gensini score due to incomplete longitudinal data in this retrospective cohort. These limitations highlight the need for future prospective, multi-centre studies with more comprehensive data collection to further explore these relationships; (3) the major goal of this experiment was to determine whether there was a link between a single baseline measurement of exposures and variables and mortality. We did not examine how risk factors or therapies changed over time; thus, we were unable to draw any inferences about the association between longitudinal risk factor control and clinical outcomes; (4) limitations are imposed by the retrospective character of this research, particularly the significant number of patients who were eliminated due to incomplete data. Therefore, our results should be interpreted with caution and need to be further verified by a prospective, multicentre study.

In conclusion, our study examined risk factors for all-cause and cardiovascular mortality in patients with coronary heart disease and diabetes. The individual’s frailty, life expectancy, diabetes duration and coronary stenosis should be considered in the decision of glycaemic control.

Ethics Statement

Registry and the registration no. of the study/trial: X2Y202352, on 2022/05/02.

Approval of the research protocol: The protocol for this research project was approved by the Ethics Committee of the Second Affiliated Hospital of Xi’an Medical University. All available data were completely anonymous with no personal information.

Informed Consent

The participants were telephoned with their consent and informed of the purpose of the study. The verbal consent was recorded by audio recording. The reasons are as follows: 1. This study is a retrospective study and all data are anonymized during the analysis process, which does not involve the privacy of patients; 2. This study has a long time span and retrospective analysis of hospitalized patients 10 years ago. Some patients died or were unable to sign informed consent due to complications during the follow-up. All specimens involving participants were approved by the Ethics Committee of the Institutional Review Board of The Second Affiliated Hospital of Xi’an Medical University (No. X2Y202352).

Funding

This study was funded by grants from National Natural Science Foundation of China (No. 82470388, 82300407), the Nature Foundation of Xi’an Administration of Science &Technology (No. 22YXYJ0157), National Postdoctoral Researcher Funding Program (GZC20233580), Health Science and Technology Innovation Capacity Improvement Program of Shaanxi Province (No. 2024TD-09, 2024JC-YBQN-0813).

Disclosure

The authors declared no conflicts of interest concerning the authorship or publication of this article.

References

1. Younossi ZM, Golabi P, Price JK, et al. The global epidemiology of nonalcoholic fatty liver disease and nonalcoholic steatohepatitis among patients with type 2 diabetes. Clin Gastroenterol Hepatol. 2024;22(10):1999–2010.e8. doi:10.1016/j.cgh.2024.03.006

2. Sun H, Saeedi P, Karuranga S, et al. IDF diabetes atlas: global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res Clin Pract. 2022;183:109119. doi:10.1016/j.diabres.2021.109119

3. Wong ND, Sattar N. Cardiovascular risk in diabetes mellitus: epidemiology, assessment and prevention. Nat Rev Cardiol. 2023;20(10):685–695. doi:10.1038/s41569-023-00877-z

4. Barnes JA, Eid MA, Creager MA, et al. Epidemiology and risk of amputation in patients with diabetes mellitus and peripheral artery disease. Arterioscler Thromb Vasc Biol. 2020;40(8):1808–1817. doi:10.1161/ATVBAHA.120.314595

5. Rajlic S, Treede H, Münzel T, Daiber A, Duerr GD. Early detection is the best prevention-characterization of oxidative stress in diabetes mellitus and its consequences on the cardiovascular system. Cells. 2023;12(4):583. doi:10.3390/cells12040583

6. Fishman SL, Sonmez H, Basman C, Singh V, Poretsky L. The role of advanced glycation end-products in the development of coronary artery disease in patients with and without diabetes mellitus: a review. Mol Med. 2018;24(1):59. doi:10.1186/s10020-018-0060-3

7. Mei Z, Xu L, Huang Q, et al. Metabonomic biomarkers of plaque burden and instability in patients with coronary atherosclerotic disease after moderate lipid-lowering therapy. J Am Heart Assoc. 2024;13(24):e036906. doi:10.1161/JAHA.124.036906

8. Riddle MC, Karl DM. Individualizing targets and tactics for high-risk patients with type 2 diabetes: practical lessons from ACCORD and other cardiovascular trials. Diabetes Care. 2012;35(10):2100–2107. doi:10.2337/dc12-0650

9. Hua X, Lung TWC, Woodward M, et al. Self-rated health scores predict mortality among people with type 2 diabetes differently across three different country groupings: findings from the ADVANCE and ADVANCE-ON trials. Diabet Med. 2020;37(8):1379–1385. doi:10.1111/dme.14237

10. Umphonsathien M, Rattanasian P, Lokattachariya S, et al. Effects of intermittent very-low calorie diet on glycemic control and cardiovascular risk factors in obese patients with type 2 diabetes mellitus: a randomized controlled trial. J Diabetes Investig. 2022;13(1):156–166. doi:10.1111/jdi.13619

11. Kloecker DE, Davies MJ, Pitocco D, et al. Intensive glucose control and recurrent cardiovascular events: 14-year follow-up investigation of the ACCORDION study. Diabetes Metab Res Rev. 2023;39(5):e3634. doi:10.1002/dmrr.3634

12. Marx N, Davies MJ, Grant PJ, et al. Guideline recommendations and the positioning of newer drugs in type 2 diabetes care. Lancet Diabetes Endocrinol. 2021;9(1):46–52. doi:10.1016/S2213-8587(20)30343-0 Erratum in: Lancet Diabetes Endocrinol. 2021 Jan;9(1):e1. doi: 10.1016/S2213-8587(20)30407-1.

13. Golub IS, Termeie OG, Kristo S, et al. Major global coronary artery calcium guidelines. JACC Cardiovasc Imaging. 2023;16(1):98–117. doi:10.1016/j.jcmg.2022.06.018

14. Shi K, Zhang G, Fu H, et al. Glycemic control and clinical outcomes in diabetic patients with heart failure and reduced ejection fraction: insight from ventricular remodeling using cardiac MRI. Cardiovasc Diabetol. 2024;23(1):148. doi:10.1186/s12933-024-02243-w

15. Yang T, Fu P, Chen J, et al. Increased risk of adverse cardiovascular events by strict glycemic control after percutaneous coronary intervention (HbA1c < 6.5% at 2 years) in type 2 diabetes mellitus combined with acute coronary syndrome: a 5-years follow-up study. Curr Med Res Opin. 2021;37(9):1517–1528. doi:10.1080/03007995.2021.1947219

16. Kavanagh ME, Back S, Chen V, et al. The portfolio diet and HbA1c in adults living with type 2 diabetes mellitus: a patient-level pooled analysis of two randomized dietary trials. Nutrients. 2024;16(17):2817. doi:10.3390/nu16172817

17. von Elm E, Altman DG, Egger M, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med. 2007;147(8):573–577. doi:10.7326/0003-4819-147-8-200710160-00010

18. Wang KY, Zheng YY, Wu TT, et al. Predictive value of gensini score in the long-term outcomes of patients with coronary artery disease who underwent PCI. Front Cardiovasc Med. 2022;8:778615. doi:10.3389/fcvm.2021.778615

19. El-Battrawy I, Demmer J, Abumayyaleh M, et al. The impact of sacubitril/valsartan on outcome in patients suffering from heart failure with a concomitant diabetes mellitus. ESC Heart Fail. 2023;10(2):943–954. doi:10.1002/ehf2.14239

20. Peeters WM, Gram M, Dias GJ, et al. Changes to insulin sensitivity in glucose clearance systems and redox following dietary supplementation with a novel cysteine-rich protein: a pilot randomized controlled trial in humans with type-2 diabetes. Redox Biol. 2023;67:102918. doi:10.1016/j.redox.2023.102918

21. Reynolds HR, Shaw LJ, Min JK, et al. Outcomes in the ISCHEMIA trial based on coronary artery disease and ischemia severity. Circulation. 2021;144(13):1024–1038. doi:10.1161/CIRCULATIONAHA.120.049755

22. Einarson TR, Acs A, Ludwig C, et al. Prevalence of cardiovascular disease in type 2 diabetes: a systematic literature review of scientific evidence from across the world in 2007-2017. Cardiovasc Diabetol. 2018;17(1):83. doi:10.1186/s12933-018-0728-6

23. Herrington WG, Alegre-Díaz J, Wade R, et al. Effect of diabetes duration and glycaemic control on 14-year cause-specific mortality in Mexican adults: a blood-based prospective cohort study. Lancet Diabetes Endocrinol. 2018;6(6):455–463. doi:10.1016/S2213-8587(18)30050-0

24. Losito A, Nunzi E, Pittavini L, et al. Cardiovascular morbidity and long term mortality associated with in hospital small increases of serum creatinine. J Nephrol. 2018;31(1):71–77. doi:10.1007/s40620-017-0413-y

25. Filippatos G, Anker SD, August P, et al. Finerenone and effects on mortality in chronic kidney disease and type 2 diabetes: a FIDELITY analysis. Eur Heart J Cardiovasc Pharmacother. 2023;9(2):183–191. doi:10.1093/ehjcvp/pvad001

26. Alicic RZ, Rooney MT, Tuttle KR. Diabetic kidney disease: challenges, progress, and possibilities. Clin J Am Soc Nephrol. 2017;12(12):2032–2045. doi:10.2215/CJN.11491116

27. Jain V, Ghosh RK, Bandyopadhyay D, et al. Serum bilirubin and coronary artery disease: intricate relationship, pathophysiology, and recent evidence. Curr Probl Cardiol. 2021;46(3):100431. doi:10.1016/j.cpcardiol.2019.06.003

28. Nano J, Muka T, Cepeda M, et al. Association of circulating total bilirubin with the metabolic syndrome and type 2 diabetes: a systematic review and meta-analysis of observational evidence. Diabetes Metab. 2016;42(6):389–397. doi:10.1016/j.diabet.2016.06.002

29. Wang J, Wu X, Li Y, et al. Serum bilirubin concentrations and incident coronary heart disease risk among patients with type 2 diabetes: the Dongfeng-Tongji cohort. Acta Diabetol. 2017;54(3):257–264. doi:10.1007/s00592-016-0946-x

30. American Diabetes Association. Older adults: standards of medical care in diabetes-2018. Diabetes Care. 2018;41(Suppl 1):S119–S125. doi:10.2337/dc18-S011.

31. Salehidoost R, Mansouri A, Amini M, et al. Diabetes and all-cause mortality, a 18-year follow-up study. Sci Rep. 2020;10(1):3183. doi:10.1038/s41598-020-60142-y

32. Christiaens A, Henrard S, Zerah L, et al. Individualisation of glycaemic management in older people with type 2 diabetes: a systematic review of clinical practice guidelines recommendations. Age Ageing. 2021;50(6):1935–1942. doi:10.1093/ageing/afab157

33. Imran SA, Agarwal G, Bajaj HS, et al. Targets for glycemic control. Can J Diabetes. 2018;42(Suppl 1):S42–S46. doi:10.1016/j.jcjd.2017.10.030