Impact of atherosclerotic cardiovascular disease on mortality in ICU sepsis patients: a retrospective MIMIC-IV cohort study Authors: Yu-Cheng Chang, Min-I Su, Wen-Pin Hu, En-Chih Liao Published: 2025 ABSTRACT Abstract AbstractObjectiveEvaluation of the impact of atherosclerotic cardiovascular disease (ASCVD), particularly polyvascular disease, on 30-day and 365-day mortality among intensive care unit (ICU) patients diagnosed with sepsis.DesignRetrospective cohort study.SettingThis study was conducted using data from the Medical Information Mart for Intensive Care IV, a publicly available critical care database from a tertiary care hospital in the US.ParticipantsA total of 7895 adult ICU patients with sepsis were included between 2008 and 2022. Among them, 3314 (41.97%) had ASCVD, defined by International Classification of Diseases (ICD) codes for coronary artery disease, peripheral artery disease or stroke.Outcome measuresThe primary outcomes were all-cause mortality at 30 and 365 days following ICU admission.ResultASCVD patients with a greater number of disease beds tend to be male, older, have lower body mass index, heart rates, mean blood pressure, respiratory rate and temperature, compared with non-ASCVD patients. Before propensity-score matching, patients with ASCVD (3314) had significantly higher 30-day (29.21% vs 24.14%, p<0.001) and 365-day mortality (30.81% vs 25.67%, p<0.001) compared with those without ASCVD (4581). Multivariate Cox regression analysis identified ASCVD as an independent predictor of 30-day (HR: 1.166, 95% CI 1.046 to 1.300, p=0.006), indicating a 16.6% higher risk of death within 30 days. For 365-day mortality, ASCVD patients had an HR of 1.168 (95% CI 1.051 to 1.298, p=0.004), reflecting a 16.8% increased risk of death within a year. Mortality risk increased incrementally with the number of involved vascular beds, especially in patients with three disease beds who exhibited the highest risk. Our findings reveal that in sepsis patients with ASCVD as a continuous variable, each additional bed increased the 30-day and 365-day mortality risk by 13.1% and 13.9%. ASCVD significantly increases the mortality risk in ICU patients with sepsis.ConclusionIntegrated management of sepsis and ASCVD—using targeted interventions such as tailored pharmacotherapy and risk factor modification—may reduce mortality and morbidity in this high-risk population.Keywords: Sepsis, Intensive Care Units, Cardiovascular Disease STRENGTHS AND LIMITATIONS OF THIS STUDY This study used a large, high-quality critical care database (Medical Information Mart for Intensive Care IV) to evaluate sepsis outcomes. This study used a large, high-quality critical care database (Medical Information Mart for Intensive Care IV) to evaluate sepsis outcomes. The severity of atherosclerotic cardiovascular disease (ASCVD) was operationalised by the number of affected vascular beds, allowing consistent categorisation of cardiovascular disease burden in all patients. The severity of atherosclerotic cardiovascular disease (ASCVD) was operationalised by the number of affected vascular beds, allowing consistent categorisation of cardiovascular disease burden in all patients. Propensity-score matching minimised confounding in comparisons between ASCVD and non-ASCVD patients. Propensity-score matching minimised confounding in comparisons between ASCVD and non-ASCVD patients. Biomarker data (eg, troponin, C-reactive protein [CRP]) were not included, limiting insights into the inflammatory-cardiovascular link. Biomarker data (eg, troponin, C-reactive protein [CRP]) were not included, limiting insights into the inflammatory-cardiovascular link. As a single-centre retrospective analysis, generalisability to other populations may be limited. As a single-centre retrospective analysis, generalisability to other populations may be limited. Introduction Sepsis, a systemic inflammatory response to infection, remains a major challenge in critical care due to its high mortality and complex pathophysiology. Comorbidities such as diabetes, hypertension and chronic kidney disease exacerbate organ dysfunction and mortality risk.1 2 Among them, atherosclerotic cardiovascular disease (ASCVD) plays a distinct role by compromising vascular integrity through pre-existing arterial damage, thereby intensifying the haemodynamic and inflammatory disturbances characteristic of sepsis. Sepsis triggers a systemic inflammatory response that accelerates endothelial dysfunction and atherothrombosis, heightening the risk of myocardial infarction, stroke and peripheral arterial disease in patients with ASCVD. Sepsis-related endothelial injury and coagulation abnormalities (eg, disseminated intravascular coagulation and microvascular thrombosis) further worsen cardiovascular outcomes. Atherosclerosis, characterised by lipid accumulation, endothelial dysfunction and chronic inflammation, leads to plaque formation and thrombosis that cause coronary, peripheral and cerebrovascular diseases. As a major global cause of morbidity and mortality, ASCVD results in significant long-term disability and demands comprehensive strategies for prevention, early detection and effective management to reduce its global burden.3 4 Polyvascular disease (PVD), defined as atherosclerosis involving multiple vascular beds, denotes a more severe and systemic form of ASCVD associated with microvascular dysfunction, reduced compensatory capacity and increased mortality.5 As the number of affected vascular territories rises, so does the risk of myocardial infarction, stroke and cardiovascular death, reflecting a more aggressive disease phenotype. Despite advances in cardiovascular care, PVD management remains complex, requiring integrated systemic therapy, long-term prevention and close monitoring.6,8 Patients with sepsis commonly present with diabetes, hypertension and chronic kidney disease, which weaken immune defences, heighten infection susceptibility and accelerate organ failure. These comorbidities increase mortality and complicate clinical management, requiring individualised treatment strategies. Compared with these conditions, ASCVD poses additional risks as pre-existing arterial lesions and endothelial dysfunction predispose patients to severe vascular and haemodynamic complications during sepsis. The synergistic effects of chronic disease burden, dysregulated inflammation and vascular injury may be underestimated by current scoring systems, underscoring the need for more holistic assessment and targeted interventions to optimise outcomes and improve survival.9 Biomarkers are essential for predicting outcomes and guiding management in patients with sepsis and ASCVD. Indicators such as Procalcitonin (PCT), CRP and troponin support early detection, risk stratification and personalised treatment. Incorporating these biomarkers into clinical practice enhances diagnostic accuracy, enables timely interventions and improves patient care.10 11 Managing patients with concurrent sepsis and ASCVD requires an integrated approach that balances infection control with cardiovascular support while considering drug interactions and individualised treatment.12 13 Patients with PVD may experience a more severe course, as sepsis combined with widespread atherosclerosis can precipitate acute cardiovascular events and multiorgan failure. The interaction between sepsis and ASCVD creates a synergistic inflammatory and vascular burden, yet the impact of multi-bed ASCVD on sepsis outcomes remains underexplored.14 15 Defining this relationship is crucial for improving risk stratification and guiding future targeted interventions. Although inflammation–atherosclerosis interactions are well established in acute coronary syndromes and stroke, the impact of multi-bed ASCVD in sepsis remains poorly defined. Sepsis-induced inflammation can exacerbate atherosclerosis across multiple vascular territories, amplifying vascular injury and mortality risk. This interaction distinguishes ASCVD from other comorbidities, as it simultaneously intensifies systemic inflammation and vascular dysfunction. Understanding how PVD affects sepsis outcomes is essential for improving prognosis and guiding targeted therapeutic strategies.16 17 This study investigated the impact of ASCVD on outcomes in critically ill patients with sepsis using the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. We examined how atherosclerosis across multiple vascular beds influences organ dysfunction, ICU course and mortality, and whether disease extent correlates with worse outcomes and higher resource utilisation. The findings aim to improve risk stratification and guide targeted management strategies for sepsis patients with PVD. Methods Data source We conducted a retrospective cohort study using data from the MIMIC-IV database,18 19 covering the period from 2008 to 2022. The MIMIC-IV database is a comprehensive, publicly available critical care dataset that includes information on 3 64 627 patients with 546 028 admissions and 94 458 ICU stays, all of whom were admitted to the intensive care units (ICUs) at the Beth Israel Deaconess Medical Centre (BIDMC) in Boston, Massachusetts.18 19 The dataset comprises de-identified patient information, including demographic data, vital signs, laboratory test results, medications and clinical outcomes, making it an invaluable resource for analysing trends and outcomes in critical care.20 The use of the MIMIC-IV database allowed for a robust and extensive analysis of a diverse patient population over a 14-year period.20 This enabled us to explore patterns and correlations between various clinical factors, including the presence of ASCVD and sepsis, and their impact on patient outcomes. Additionally, the database includes out-of-hospital mortality data, enriching the scope of long-term follow-up. Following the completion of web-based training and the successful completion of the Protecting Human Research Participants examination, author Yucheng Chang (Record ID: 9780592) obtained the necessary approval to access and extract data from the MIMIC-IV database (https://github.com/MIT-LCP/mimic-iv).19 Study population This study was conducted in compliance with the ethical guidelines and requirements set forth by the Institutional Review Board (IRB) of BIDMC and the database administrators. This MIMIC-IV cohort study was approved by the Institutional Review Board of MacKay Memorial Hospital (IRB No. 24MMHIS460e). The requirement for informed consent was waived because the study used retrospective, de-identified clinical data. In the MIMIC-IV database (V.3.1), we initially identified 41 296 patients diagnosed with sepsis, defined by a Sequential Organ Failure Assessment (SOFA) score >2 and the presence of a suspicious infection, among a total of 94 458 ICU admissions. The SOFA score is a clinical tool used to evaluate the severity of organ dysfunction in critically ill patients, including those with sepsis. It assesses six organ systems: (1) Respiratory system: PaO₂/FiO₂ ratio (mm Hg), (2) Coagulation: platelet count (×10⁹/L), (3) Liver function: bilirubin level (mg/dL), (4) Cardiovascular system: hypotension and vasopressor requirements, (5) Central nervous system: Glasgow Coma Scale (GCS) score and (6) Renal function: creatinine level (mg/dL) or urine output (mL/day). Each system is scored from 0 (normal function) to 4 (severe dysfunction), and the total SOFA score is obtained by summing the individual scores of all six systems. Higher total scores reflect greater organ dysfunction and an increased risk of mortality. To ensure the accuracy and relevance of our cohort, we applied several exclusion criteria. First, we excluded individuals under 18 years of age to focus on the adult population. We also excluded patients who had an ICU stay of less than 6 hours, as such short stays might not reflect a full critical care episode or provide adequate clinical data for meaningful analysis. In addition, patients with repeated ICU admissions were removed to avoid duplication and ensure that each patient represented a unique case. Finally, we eliminated patients lacking complete hospitalisation records, as missing data could skew the results and limit the robustness of our findings. This refined sample enabled us to focus on a population with consistent clinical data, offering a clearer understanding of sepsis outcomes and the role of ASCVD in these critically ill patients. Coding of diseases and symptoms Patients in this study were classified into two groups: non-ASCVD and ASCVD (having one or more ASCVDs). The definition and severity of ASCVD was based on the presence of coronary artery disease (CAD), peripheral artery disease (PAD) and stroke using the relevant ICD-9 and ICD-10 codes. A fully annotated list of all included ICD-9 and ICD-10 codes has been provided in the online supplemental table 1). Patients with CAD were identified using the ICD-10 codes I20~I22 and I24~I25, and the corresponding ICD-9 codes 410~414. For those diagnosed with PAD, we applied the ICD-10 codes I740~I745, I748, I7502, I771, E0852, E0952, E105, E115 and E135, along with ICD-9 codes 43300, 43310, 43320, 43330, 43380, 43390, 43400, 43410, 43490, 440, 4439, 44381 and 44389. To identify patients with a history of stroke, we used the ICD-10 codes I63 and G45, and the ICD-9 codes 43311, 43321, 43331, 43381, 43391, 43401, 43411, 43491 and 435. In MIMIC-IV, sepsis is most identified using the Sepsis-3 criteria,21 which rely on clinical indicators—specifically, a suspected infection combined with a SOFA score increase of 2 or more—rather than solely on ICD codes. ICD codes are used to define sepsis, ICD-9 codes: 995.91 (Sepsis) and 995.92 (Severe sepsis); ICD-10 codes: A40.x (Streptococcal sepsis) and A41.x (Other sepsis). This comprehensive coding system enabled us to accurately classify patients based on their ASCVD status, ensuring we included those with significant atherosclerotic disease affecting multiple vascular territories. The number of disease beds represented the extent of ASCVD involvement across three vascular territories—coronary, peripheral and cerebrovascular. Each affected bed (CAD, PAD or stroke) counted as one point, yielding a total score from 0 to 3. Patients were thus classified as having 0, 1, 2 or 3 disease beds, with higher numbers indicating greater polyvascular burden. Clinical parameters We extracted a range of patient parameters, including sex, age, body mass index (BMI), heart rate, mean blood pressure (MBP), respiratory rate, temperature, SpO2, glucose, lactate, BUN, total calcium, sodium, potassium and eGFR. BMI data were obtained from the admission body weight and height recorded in the chartevents table of the MIMIC-IV database, as documented by nursing staff during ICU admission. These measurements were used to calculate BMI at baseline. Additionally, we collected data on haemoglobin, platelet count, WBC and ALT. Comorbid conditions were also noted, such as congestive heart failure, chronic pulmonary disease, smoking status, hypertension, dyslipidaemia and diabetes. To assess the severity and prognosis of the patients, we included clinical scores such as the SOFA, GCS, Simplified Acute Physiology Score II (SAPS II) and Logistic Organ Dysfunction System (LODS). Although cardiac biomarkers (eg, troponin, BNP), inflammatory biomarkers (eg, CRP, PCT) and echocardiographic data were available in the MIMIC-IV database, they were not included in the current analysis, due to significant data sparseness. Statistics The study population is stratified by the number of disease beds (None, 1, 2 or 3) and we conducted descriptive statistics for both continuous and binary variables. Continuous variables were presented as means and SD, while binary variables were expressed as numbers and percentages. To evaluate differences between groups, we used independent t-tests for continuous variables and chi-squared tests for binary variables, ensuring appropriate statistical comparisons based on the type of data. Kaplan-Meier curves were used and the log-rank test was performed to compare 30-day and 365-day mortality differences between non-ASCVD, 1, 2 and 3 disease bed groups. To balance the potential confounders, we applied 1:1 propensity-score matching (PSM) to match the baseline characteristics between the two groups (ASCVD and non-ASCVD) with a calliper width of 0.2.22 The Cox proportional-hazards model was employed to investigate the relationships between covariates and outcomes after matching. Significant variables were subsequently incorporated into multivariate Cox regression models to evaluate their individual associations with outcomes, with results presented as HRs, 95% CIs and corresponding p values. Three progressively adjusted models were built: Model 1 adjusted for basic demographic variables (sex, age, BMI); Model 2 included additional clinical factors such as heart rate, respiratory rate, blood pressure and laboratory values and Model 3 further incorporated comorbidities, including congestive heart failure, chronic pulmonary disease, hypertension, diabetes and others. The potential multicollinearity in our multivariate Cox proportional-hazards model was assessed using a variance inflation factor (VIF) analysis for all covariates included in the model. Results showed that the VIF values for all variables were below 5, indicating low multicollinearity and confirming the stability and reliability of our model. The adjusted HRs (aHRs) demonstrate the relative risk of mortality for different groups (eg, ASCVD vs Non-ASCVD, number of disease beds as continuous variables), with 95% CIs and p values indicating statistical significance. All p values were two-tailed, with a significance level set at p≤0.05. We conducted all statistical analyses using the SPSS software (V.20.0) and R software (V.4.2.3), ensuring the rigorous application of both parametric and survival analysis techniques. Mortality The primary outcome of this study was 30-day and 365-day mortality in ICU patients with sepsis, comparing those with and without ASCVDs. These outcomes were used to assess the impact of ASCVD on short-term and long-term mortality in critically ill sepsis patients. Results In the MIMIC-IV database (V.3.1), out of 94 458 ICU admissions, 41 296 patients were diagnosed with sepsis (defined as having a SOFA score >2 and a suspected infection). After excluding patients under 18 years of age, those with ICU stays of less than 6 hours and those with repeated admissions, 29 320 patients remained. Further exclusion of patients lacking hospitalisation records resulted in a final cohort of 7895 patients being included in the study (figure 1). After applying these criteria, the final study cohort comprised 7895 patients (figure 1). Among these 7895 patients were enrolled in this study, 3314 (41.97%) having one or more ASCVDs. Table 1 demonstrated the baseline characteristics between the groups. The overall mean age of the cohort was 63.89±16.41 years and the majority were men (59.16%). ASCVD patients with a greater number of disease beds tend to be male, older, have lower BMIs, heart rates, MBP, respiratory rate and temperature, compared with patients in the non-ASCVD group. Patients with ASCVD exhibit more pronounced laboratory abnormalities, including elevated total calcium and potassium levels, as well as reduced estimated glomerular filtration rate (eGFR). Chronic comorbidities such as congestive heart failure, chronic pulmonary disease, hypertension and diabetes mellitus are more common in groups with one or more disease beds. Flowchart of this research cohort. This flowchart illustrates the stepwise process of selecting the final cohort of 7895 sepsis patients from an initial 94 458 ICU admissions in the MIMIC-IV database, applying various exclusion criteria. ASCVD, atherosclerotic cardiovascular disease; ICU, intensive care unit; MIMIC-IV, Medical Information Mart for Intensive Care IV; SOFA, Sequential Organ Failure Assessment. Table 1Baseline characteristicsCharacteristicAll patients (n=7895)Number of disease bedsP valueNone123Mean (SD)/n(%)(n=4581)(n=2432)(n=792)(n=90)Sex (male)4671 (59.16)2619 (57.17)1507 (61.97)493 (62.25)52 (57.78)<0.001Age (years)63.89 (16.19)59.42 (16.68)69.43 (13.49)71.76 (12.19)72.78 (11.89)<0.001BMI (kg/m2)29.53 (8.38)29.8 (8.88)29.48 (7.85)28.3 (6.77)27.82 (7.63)<0.001Heart rate (beats per minute)95.03 (22)97.87 (22.04)91.47 (21.35)90.24 (21.38)88.63 (20.62)<0.001MBP (mm Hg)83.25 (20.26)83.37 (20.19)83.23 (20.28)82.73 (20.92)82.14 (17.2)0.815Respiratory rate (insp/min)21.42 (6.48)21.76 (6.68)21.06 (6.15)20.55 (6.13)20.76 (6.85) <0.001 Temperature (°C)36.72 (1.09)36.77 (1.11)36.66 (1.08)36.58 (1.05)36.68 (0.91) <0.001 SpO2 (%)96.27 (5.1)96.24 (5.2)96.26 (4.9)96.37 (5.21)96.79 (4.06)0.719Glucose (mg/dL)166.85 (94.2)157.92 (87.61)177.04 (102.57)185.76 (97.65)180.23 (98.33) <0.001 Lactate (mmol/L)2.87 (2.6)2.94 (2.66)2.69 (2.41)2.96 (2.82)2.79 (2.18) 0.001 Total calcium (mg/dL)8.27 (1.07)8.21 (1.13)8.34 (0.98)8.37 (0.94)8.37 (0.9) <0.001 Sodium (mEq/L)137.86 (6.31)137.81 (6.59)138.11 (5.99)137.39 (5.54)137.4 (6.7) 0.030 Potassium (mEq/L)4.46 (1)4.41 (1)4.48 (0.97)4.63 (1.03)4.73 (0.9) <0.001 eGFR (mL/min)61.23 (46.76)66.75 (50.85)55.68 (39.68)48 (37.58)46.75 (33.59) <0.001 Haemoglobin (g/L)10.96 (2.62)10.96 (2.66)11.1 (2.57)10.69 (2.53)10.15 (2.73) <0.001 Platelet (k/µL)212.21 (125.55)208.94 (127.66)214.55 (122.49)222.2 (123.01)227.13 (115.46) 0.017 WBC (k/µL)14.42 (12.16)14.57 (13.2)14.27 (11.37)14.02 (7.57)14.65 (9.73)0.579ALT (IU/L)179.74 (670.6)187.73 (721.42)176.97 (634.19)141.14 (460.13)187.24 (469.84)0.344Congestive heart failure2661 (33.7)917 (20.02)1196 (49.18)497 (62.75)51 (56.67) <0.001 Chronic pulmonary disease2083 (26.38)1076 (23.49)726 (29.85)260 (32.83)21 (23.33) <0.001 Smoking707 (8.96)422 (9.21)182 (7.48)92 (11.62)11 (12.22) 0.002 Hypertension5150 (65.23)2565 (55.99)1827 (75.12)681 (85.98)77 (85.56) <0.001 Dyslipidaemia3339 (42.29)1406 (30.69)1308 (53.78)557 (70.33)68 (75.56) <0.001 Diabetes mellitus2582 (32.7)1132 (24.71)956 (39.31)443 (55.93)51 (56.67) <0.001 SOFA score8.09 (3.88)8.15 (3.98)7.97 (3.83)8.15 (3.52)8.17 (3.27)0.340SAPS II score46.01 (15.61)44.55 (15.92)47.75 (15.08)48.78 (14.48)49.01 (14.8) <0.001 LODS score7.13 (3.09)6.98 (3.1)7.28 (3.08)7.52 (3.06)7.54 (2.98) <0.001 Length of ICU stay (day)8.78 (9.21)9.07 (9.66)8.34 (8.45)8.63 (9.07)7.45 (6.16) 0.007 Length of hospital stay (day)17.67 (17.31)18.33 (18.13)16.34 (15.78)17.8 (16.46)18.5 (19.85) <0.001 Followed up (day)318 (714.34)357.54 (783.15)269.46 (610.97)250.37 (572.05)212.67 (630.57) <0.001 30 days mortality2074 (26.27)1106 (24.14)689 (28.33)245 (30.93)34 (37.78) <0.001 365 days mortality2197 (27.83)1176 (25.67)725 (29.81)258 (32.58)38 (42.22) <0.001 The cohort’s mean age was 63.9±16.4 years, with 59.2% men. Patients with atherosclerotic cardiovascular disease were generally older, more often male and had lower BMI, vital signs and eGFR but higher calcium and potassium levels. Chronic comorbidities—including heart failure, pulmonary disease, hypertension and diabetes—were more prevalent with increasing numbers of disease beds.Bold font display significant difference.ALT, Alanine Aminotransferase; BMI, Body mass index; eGFR, estimated Glomerular Filtration Rate; LODS, Logistic Organ Dysfunction Score; MBP, Mean blood pressure; SAPS II, Simplified Acute Physiology Score II; SOFA, Sequential Organ Failure Assessment; SpO2, O2 saturation pulse oximetry; WBC, White Blood Cell. Baseline characteristics The cohort’s mean age was 63.9±16.4 years, with 59.2% men. Patients with atherosclerotic cardiovascular disease were generally older, more often male and had lower BMI, vital signs and eGFR but higher calcium and potassium levels. Chronic comorbidities—including heart failure, pulmonary disease, hypertension and diabetes—were more prevalent with increasing numbers of disease beds. Bold font display significant difference. ALT, Alanine Aminotransferase; BMI, Body mass index; eGFR, estimated Glomerular Filtration Rate; LODS, Logistic Organ Dysfunction Score; MBP, Mean blood pressure; SAPS II, Simplified Acute Physiology Score II; SOFA, Sequential Organ Failure Assessment; SpO2, O2 saturation pulse oximetry; WBC, White Blood Cell. Regarding sex, we have conducted subgroup analyses to explore potential sex-specific differences in the outcomes. The proportion of males progressively increases from the ‘None’ disease bed group (57.17%) to the ‘1-bed’ group (61.97%) and the ‘2-bed’ group (62.25%). However, it slightly decreases in the ‘3-bed’ group (57.78%) (table 1). The results indicate that this difference in male proportions across the groups is statistically significant (p<0.001), suggesting an association between male sex and the presence of ASCVD. The mean LODS score for all ASCVD patients was 7.13 (± 3.09), increasing progressively with the number of disease beds, from 6.98 in patients with non-ASCVD to 7.54 in those with three ASCVD disease beds. Similarly, the SAPS II score showed a consistent rise, with an overall mean of ASCVD patients of 46.01 (± 15.61), ranging from 44.55 in patients with non-ASCVD to 49.01 in those with three disease beds. Mortality rates demonstrated a parallel trend; 30-day mortality was 26.27% for all ASCVD patients but increased from 24.14% in non-ASCVD to 37.78% in those ASCVD patients with three disease beds. The 365-day mortality followed a similar pattern, rising from 25.67% (non-ASCVD) to 42.22% (ASCVD with three disease beds). The Kaplan-Meier survival curves (figures2 3) highlight the impact of the number of disease beds (1, 2 or 3) on patient survival. In both figures, survival probabilities decline significantly with increasing disease burden. Patients with no disease beds show the highest survival rates, while those with one or two disease beds exhibit progressively poorer survival. The group with three disease beds experiences the steepest decline in survival, with the curves diverging sharply early on, indicating a significant association between higher disease burden and mortality (p<0.001). Kaplan-Meier curves of 30-day mortality risk. The survival curve shows that sepsis patients with atherosclerotic cardiovascular disease (ASCVD) have a higher 30-day mortality risk compared to those without ASCVD, with a statistically significant difference (p<0.001). Kaplan-Meier curves of 365-day mortality risk. This curve highlights the increased long-term (365-day) mortality risk in sepsis patients with atherosclerotic cardiovascular disease, also showing a significant difference between the groups (p<0.001). We performed PSM for ASCVD and non-ASCVD groups (table 2) to minimise the confounders in this retrospective study. Before PSM, significant differences were observed across several variables, such as sex, age, BMI, heart rate, respiratory rate, temperature, lactate, glucose levels and total calcium, potassium, eGFR and platelet. Comorbidities such as congestive heart failure, chronic pulmonary disease, hypertension, dyslipidaemia and diabetes mellitus were more prevalent in the ASCVD group. The SAPS II score was higher in the ASCVD group (48.03±14.94 vs 44.55±15.92) and the LODS score was also elevated in the ASCVD group (7.35±3.07 vs 6.98±3.10), compared with the non-ASCVD group, with a statistically significant difference (p<0.001). Mortality outcomes were also significantly different, with 30-day mortality at 24.14% for non-ASCVD versus 29.21% for ASCVD (p<0.001) and 365-day mortality at 25.67% for non-ASCVD versus 30.81% for ASCVD (p<0.001). Table 2Propensity score matchingCharacteristicBefore matching (n=7895)After matching (n=4916)Non-ASCVD(n=4581)ASCVD(n=3314)P valueSMDNon-ASCVD(n=2458)ASCVD(n=2458)P valueSMDMean (SD)/n(%)Sex (male)2619 (57.17)2052 (61.92) <0.001 −0.101453 (59.11)1499 (60.98)0.095−0.04Age (years)59.42 (16.68)70.08 (13.19) <0.001 0.8167.49 (14.11)68.06 (13.56)0.1480.04BMI (kg/m2)29.8 (8.88)29.16 (7.61) <0.001 −0.0929.45 (8.64)29.31 (7.93)0.555−0.02Heart rate (beats per minute)97.87 (22.04)91.1 (21.34) <0.001 −0.3294.07 (21.74)92.5 (21.82) 0.012 −0.07MBP (mm Hg)83.37 (20.19)83.08 (20.36)0.537−0.0183.08 (20.21)83.21 (20.04)0.8240.01Respiratory rate (insp/min)21.76 (6.68)20.93 (6.16) <0.001 −0.1321.39 (6.52)21.02 (6.23) 0.043 −0.06Temperature (°C)36.77 (1.11)36.64 (1.07) <0.001 −0.1236.71 (1.11)36.67 (1.08)0.146−0.04SpO2 (%)96.24 (5.2)96.3 (4.95)0.6110.0196.18 (5.38)96.35 (4.77)0.2270.04Glucose (mg/dL)157.92 (87.61)179.21 (101.34) <0.001 0.21169.79 (99.99)172.11 (95.3)0.4050.02Lactate (mmol/L)2.94 (2.66)2.76 (2.51) 0.002 −0.072.82 (2.47)2.78 (2.54)0.545−0.02Total calcium (mg/dL)8.21 (1.13)8.34 (0.97) <0.001 0.148.29 (1.1)8.3 (0.96)0.6030.02Sodium (mEq/L)137.81 (6.59)137.92 (5.91)0.4630.02137.94 (6.67)138 (5.93)0.7250.01Potassium (mEq/L)4.41 (1)4.52 (0.99) <0.001 0.114.49 (1)4.49 (0.98)0.9970.00eGFR (mL/min)66.75 (50.85)53.6 (39.18) <0.001 −0.3457.42 (40.6)56.81 (42.1)0.606−0.02Haemoglobin (g/L)10.96 (2.66)10.97 (2.58)0.7650.0111 (2.59)11.02 (2.58)0.8420.01Platelet (k/µL)208.94 (127.66)216.72 (122.45) 0.006 0.06213.84 (123.66)214.91 (125)0.7640.01WBC (k/µL)14.57 (13.2)14.22 (10.54)0.192−0.0314.37 (13.68)14.28 (11.03)0.797−0.01ALT (IU/L)187.73 (721.42)168.69 (593.16)0.199−0.03184.5 (729.47)177.76 (615.15)0.726−0.01Congestive heart failure917 (20.02)1744 (52.63) <0.001 0.65848 (34.5)983 (39.99) <0.001 0.11Chronic pulmonary disease1076 (23.49)1007 (30.39) <0.001 0.15692 (28.15)705 (28.68)0.6810.01Smoking422 (9.21)285 (8.6)0.359−0.02219 (8.91)222 (9.03)0.9200.00Hypertension2565 (55.99)2585(78) <0.001 0.531786 (72.66)1798 (73.15)0.7240.01Dyslipidaemia1406 (30.69)1933 (58.33) <0.001 0.561198 (48.74)1257 (51.14)0.0980.05Diabetes mellitus1132 (24.71)1450 (43.75) <0.001 0.38889 (36.17)941 (38.28)0.1320.04SOFA score8.15 (3.98)8.02 (3.75)0.149−0.038.04 (3.77)8.01 (3.83)0.799−0.01SAPS II score44.55 (15.92)48.03 (14.94) <0.001 0.2347.3 (15.42)47.25 (15.05)0.8940.00LODS score6.98 (3.1)7.35 (3.07) <0.001 0.127.2 (3.1)7.23 (3.1)0.7480.01Length of ICU stay (day)9.07 (9.66)8.39 (8.55) <0.001 8.67 (8.49)8.6 (8.79)0.762Length of hospital stay (day)18.33 (18.13)16.75 (16.08) <0.001 17.33 (16)17.2 (16.56)0.777Followed up (day)357.54 (783.15)263.35 (602.37) <0.001 362.64 (796.83)276.13 (615.4) <0.001 30 days mortality1106 (24.14)968 (29.21) <0.001 619 (25.18)688 (27.99) 0.028 365 days mortality1176 (25.67)1021 (30.81) <0.001 656 (26.69)728 (29.62) 0.024 Before propensity score matching (PSM), Atherosclerotic cardiovascular disease (ASCVD) patients were older, more often male, and had higher comorbidity and severity scores than non-ASCVD patients (all p<0.001). After 1:1 PSM (n=4916; SMD <0.2), baseline differences were balanced, yet mortality remained higher in the ASCVD group (30-day: 27.99% vs 25.18%, p=0.028; 365-day: 29.62% vs 26.69%, p=0.024).Bold font display significant difference.ALT, Alanine Aminotransferase; BMI, Body mass index; eGFR, estimated Glomerular Filtration Rate; LODS, Logistic Organ Dysfunction Score; MBP, Mean blood pressure; SAPS II, Simplified Acute Physiology Score II; SOFA, Sequential Organ Failure Assessment; SpO2, O2 saturation pulse oximetry; WBC, White Blood Cell. Propensity score matching Before propensity score matching (PSM), Atherosclerotic cardiovascular disease (ASCVD) patients were older, more often male, and had higher comorbidity and severity scores than non-ASCVD patients (all p<0.001). After 1:1 PSM (n=4916; SMD <0.2), baseline differences were balanced, yet mortality remained higher in the ASCVD group (30-day: 27.99% vs 25.18%, p=0.028; 365-day: 29.62% vs 26.69%, p=0.024). Bold font display significant difference. ALT, Alanine Aminotransferase; BMI, Body mass index; eGFR, estimated Glomerular Filtration Rate; LODS, Logistic Organ Dysfunction Score; MBP, Mean blood pressure; SAPS II, Simplified Acute Physiology Score II; SOFA, Sequential Organ Failure Assessment; SpO2, O2 saturation pulse oximetry; WBC, White Blood Cell. After PSM, 4916 patients remained, comprising 2458 matched pairs in both groups and the two groups were well balanced, as indicated by small standardised mean differences (SMD <0.2) across variables. Mortality outcomes post-matching remained statistically significant, with 30-day mortality at 25.18% for non-ASCVD versus 27.99% for ASCVD (p=0.028) and 365-day mortality at 26.69% for non-ASCVD versus 29.62% for ASCVD (p=0.024). Overall, PSM effectively reduced baseline differences, allowing for a more accurate comparison of outcomes between Non-ASCVD and ASCVD groups. The univariate analysis presented in online supplemental table 2A demonstrates that patients with ASCVD had a HR of 1.166 (95% CI 1.046 to 1.300, p=0.006) compared with non-ASCVD patients, indicating a 16.6% higher risk of death within 30 days. For 365-day mortality, the univariate analysis also revealed significant associations. ASCVD patients had an HR of 1.168 (95% CI 1.051 to 1.298, p=0.004), reflecting a 16.8% increased risk of death within a year (online supplemental table 2B). The Cox regression analysis for multivariate model highlights the critical impact of ASCVD status and the number of disease beds as continuous variables on 30-day and 365-day mortality outcomes. Patients with ASCVD consistently exhibited higher mortality risks compared with those without ASCVD across all models. For 30-day mortality, the aHR for ASCVD versus non-ASCVD in Model 3 was 1.187 (95% CI 1.064 to 1.320, p=0.002), indicating an 18.7% higher risk of death within 30 days (online supplemental table 2A). Similarly, for 365-day mortality, ASCVD patients had an aHR of 1.189 (95% CI 1.069 to 1.322, p=0.001), reflecting a sustained higher risk over the long term (online supplemental table 2B). When analysing the number of disease beds as a continuous variable, the aHR also demonstrated a clear and significant association with mortality. For 30-day mortality, the aHR in Model 3 was 1.131 (95% CI 1.050 to 1.218, p=0.001), suggesting that each additional disease bed increased the risk of death by 13.1%. Similarly, for 365-day mortality, the aHR was 1.139 (95% CI 1.060 to 1.224, p<0.001), underscoring a consistent increase in mortality risk with greater disease burden. When comparing 30-day mortality results between the number of disease beds (none, one, two or three), patients with one disease bed had an aHR of 1.176 (95% CI 1.046 to 1.322, p=0.007) compared with those with no disease bed, while those with two disease beds had an aHR of 1.159 (95% CI 0.964 to 1.393, p=0.116), which was not statistically significant. However, patients with three disease beds exhibited the highest risk, with an aHR of 1.896 (95% CI 1.247 to 2.883, p=0.003). A similar trend was observed for 365-day mortality. One disease bed was associated with an aHR of 1.171 (95% CI 1.045 to 1.313, p=0.007), while two disease beds had an aHR of 1.168 (95% CI 0.977 to 1.396, p=0.088), and three disease beds carried the highest risk, with an aHR of 2.048 (95% CI 1.380 to 3.041, p<0.001). The inclusion of composite severity scores (SOFA, SAPS II and LODS) in the same multivariable model may represent potential mediators on the causal pathway between ASCVD and mortality, which could lead to over-adjustment bias and attenuation of the total effect of ASCVD. To address this concern and ensure a more accurate estimation of the total effect, we have revised our analytical framework and conducted two additional step-up sensitivity models to evaluate the potential impact of these severity scores: Model 4 (Step-up inclusion of SOFA): SOFA score was sequentially added to the base adjustment model. Model 5 (Inclusion of SOFA and SAPS II): both SOFA and SAPS II scores were incorporated simultaneously to further assess model robustness. Importantly, the prognostic association between ASCVD and mortality remained statistically significant across all restructured models (Model 4 and Model 5). These consistent results indicate that the observed relationship is not solely explained by acute physiological severity, thereby reinforcing the independent prognostic role of ASCVD in this critically ill sepsis population. Additionally, the detailed results of Model 4 and Model 5 have been provided in the revised manuscript as online supplemental table 3 to allow full transparency and reproducibility of the analyses. Discussion Our study shows that sepsis-related cardiovascular dysfunction—especially in patients with multiple ASCVD conditions—increases mortality. This highlights the need for multidisciplinary interventions that balance infection control with cardiovascular support. As sepsis incidence rises in an ageing, multimorbid population, particularly among those with PVD, our findings underscore the importance of integrating critical care, cardiology and infectious disease expertise to refine risk assessment and personalise treatment, ultimately guiding future therapeutic strategies. In this retrospective cohort study of 7895 ICU sepsis patients from the MIMIC-IV database (41.97% with ASCVD), those with ASCVD had higher mortality (30-day: 29.21% vs 24.14%; 365-day: 30.81% vs 25.67%). Kaplan-Meier curves confirmed lower survival in the ASCVD group. These findings highlight that ASCVD significantly increases short- and long-term mortality, emphasising the need for early detection and integrated management of sepsis and cardiovascular comorbidities. Our multivariate Cox regression analysis identified ASCVD as an independent predictor of increased mortality at 30 days (HR: 1.166) and 365 days (HR: 1.168), with risk escalating alongside greater vascular bed involvement. Key parameters associated with higher 30-day mortality included age, respiratory rate, lactate, INR, WBC count, diabetes, SAPS II and LODS scores, whereas higher BMI, temperature, chloride and platelet counts were protective. These results underscore the compounded impact of sepsis and ASCVD on outcomes and highlight the need for integrated management strategies in this high-risk population. The interplay between sepsis and ASCVD stems from shared mechanisms such as systemic inflammation, endothelial dysfunction and coagulopathy. Sepsis worsens endothelial injury, accelerates atherosclerosis and creates a prothrombotic state, thereby increasing cardiovascular risk. In patients with ASCVD, this inflammatory and prothrombotic environment further exacerbates vascular damage, leading to complications like myocardial infarction and stroke, while reduced vascular reactivity and impaired cardiac reserve heighten the risk of organ failure and mortality. These findings underscore the need for targeted strategies to mitigate the combined impact of sepsis and ASCVD on patient outcomes. It is noteworthy that the markedly increased mortality risk observed in patients with PVD may be attributed to the cumulative burden of multiple affected vascular beds. The diffuse nature of atherosclerosis in PVD results in widespread endothelial dysfunction and impaired microvascular perfusion, which further compromises the cardiovascular system’s ability to respond to the haemodynamic challenges of sepsis. In the setting of septic shock, these patients often have a diminished compensatory reserve to manage vasodilation, capillary leak and fluid shifts, thereby experiencing significantly higher mortality. Furthermore, when comparing sepsis patients with PVD to those with other high-risk conditions such as heart failure or chronic kidney disease, the mortality risk associated with PVD appears to be of a similar magnitude. This suggests that the cumulative vascular burden in PVD could serve as an independent high-risk marker, warranting similar levels of clinical attention and aggressive management as these well-established predictors of sepsis mortality. Multiple ASCVD or PVD (atherosclerosis in two or more arterial beds) significantly worsens cardiovascular outcomes. In cardiogenic shock, PVD is linked to higher mortality and more major adverse cardiovascular events (MACE) compared with non-PVD patients.23 The cumulative atherosclerotic burden in PVD compromises haemodynamic stability and organ perfusion, leading to increased 12-month all-cause mortality—especially in patients with lower vasoactive inotropic scores. Among symptomatic PAD patients, PVD independently predicts higher long-term mortality, with the EUCLID trial showing that PAD patients with additional coronary or cerebrovascular disease face higher MACE incidence and more frequent lower-extremity revascularisations.24 These findings underscore the need for aggressive management and close monitoring of PVD to improve outcomes. Patients with PVD in septic shock face worse outcomes due to chronic cardiovascular risk factors (eg, hypertension, diabetes, dyslipidaemia, smoking) that promote a pro-inflammatory state and widespread atherosclerosis. During septic shock, this inflammation is amplified, leading to endothelial dysfunction, coagulopathy and reduced tissue perfusion, which increase the risk of multi-organ failure.25 Our study confirms that patients with multiple ASCVDs have higher mortality and a greater risk of multi-organ failure compared with those without PVD, underscoring the need for early detection and integrated therapeutic strategies targeting both inflammation and vascular dysfunction.26 Chronic inflammation and endothelial dysfunction in PVD exacerbate vascular damage, coagulopathy and impaired organ perfusion during septic shock, increasing the risk of multi-organ failure and mortality. Targeting both infection and underlying inflammation is crucial for improving outcomes and developing personalised treatments for this high-risk population. Future research should investigate the mechanisms linking ASCVD to higher mortality in sepsis, focusing on inflammation, endothelial dysfunction and coagulopathy. Cytokines (eg, IL-6, TNF-alpha) and adhesion molecules (eg, ICAM-1, VCAM-1) may drive endothelial injury, vascular inflammation and atherosclerosis progression while intensifying the systemic inflammatory response in sepsis. This interplay impairs organ perfusion and increases the risk of multi-organ failure. Additionally, shared coagulopathy may heighten thrombotic events, worsening cardiovascular outcomes.27 Understanding these pathways could guide targeted therapies to improve outcomes in patients with both sepsis and ASCVD. Identifying biomarkers to predict outcomes in sepsis patients with ASCVD could improve clinical management and prognostic accuracy. Cardiac markers like troponins and natriuretic peptides, along with sepsis markers such as procalcitonin and CRP, provide insight into cardiovascular stress and systemic inflammation.28 Troponins indicate myocardial injury and sepsis-induced cardiac dysfunction, both linked to higher mortality.29 Natriuretic peptides (BNP, NT-proBNP) reflect cardiac strain and heart failure risk, crucial in managing ASCVD during sepsis.30 Inflammatory markers like procalcitonin and CRP assess infection severity and systemic inflammation. Integrating cardiac and inflammatory biomarkers may enable personalised interventions and improve outcomes. Future research should explore this approach to optimise care for high-risk patients. Targeted therapies addressing both sepsis and ASCVD are gaining interest, with GLP-1 receptor agonists (GLP-1RAs) showing potential. Originally used to reduce cardiovascular events in diabetes, GLP-1RAs may benefit septic patients with ASCVD by modulating inflammation and improving endothelial function.31 Their anti-inflammatory properties could mitigate the systemic inflammatory response in sepsis, reducing organ dysfunction, while their endothelial benefits may counteract vascular inflammation and dysfunction linked to sepsis and atherosclerosis. This dual mechanism suggests GLP-1RAs could improve outcomes in this high-risk population. Further research is needed to evaluate their impact on mortality and cardiovascular events, guiding personalised treatments. Personalised medicine leveraging genomics and proteomics may optimise treatment for sepsis patients with ASCVD. By identifying genetic predispositions and protein expression profiles, clinicians can tailor therapies to individual responses.32 This approach uncovers mutations that increase susceptibility to severe sepsis or cardiovascular complications, enabling precision treatments that improve efficacy, reduce drug interactions and enhance survival outcomes in this high-risk population. Robust clinical trials are essential to evaluate interventions in sepsis patients with ASCVD, focusing on both survival and long-term recovery. Research on therapies like antiplatelet agents, statins and anti-inflammatory drugs is needed to address the interplay between infection, inflammation and cardiovascular disease. A multidisciplinary approach involving cardiology, critical care and infectious diseases is crucial to develop evidence-based protocols that optimise outcomes and reduce complications in this high-risk population. Our work seeks to influence clinical decisions and future research by promoting personalised, evidence-based care for critically ill patients with sepsis and ASCVD. As PVD rises in an ageing population alongside increasing sepsis incidence, tailored therapies are crucial. Our study offers insights to refine treatment protocols and explore novel interventions targeting inflammatory and cardiovascular pathways to reduce mortality and improve outcomes. By enhancing understanding of these complex interactions, we aim to advance critical care and drive precision medicine research for patients with multiple comorbidities. Previous studies have shown that multiple vascular bed involvement in cardiovascular disease correlates with higher mortality.33 However, research on this relationship in sepsis patients remains limited. Our findings reveal a similar trend, with each additional affected vascular bed increasing 30-day and 365-day mortality risk by 13.1% and 13.9%, respectively. Patients with three affected beds had the highest mortality risk (aHRs: 1.896 for 30-day, 2.048 for 365-day), both statistically significant. In contrast, those with two beds showed an increased risk that did not reach significance (p>0.05), possibly due to sample size limitations or confounders. These results highlight the need for targeted interventions and close monitoring, particularly for sepsis patients with extensive vascular involvement. This study has several limitations. Although the MIMIC-IV database provides detailed ICU-level physiological and laboratory data, it lacks certain granular information such as imaging results, behavioural factors and complete outpatient histories, which may limit analytical depth. Missing data (eg, 23.65% of ethnicity information) and selection bias from its single-centre, retrospective design may also affect generalisability. Additionally, while we included key inflammatory and cardiac biomarkers (eg, CRP, procalcitonin, troponin, CK-MB, BNP) when available, incomplete data limited their comprehensive analysis. This may have introduced variability, as these markers could provide deeper insights into the inflammatory–cardiovascular interplay; nevertheless, their clinical relevance remains clear. Elevated cardiac troponins are crucial for assessing prognosis in ASCVD patients, as they are strongly linked to adverse cardiovascular events and mortality. In PAD, higher troponin levels correlate with worse long-term outcomes,34 while in sepsis, they predict both in-hospital and long-term mortality.35 36 These findings highlight troponin’s role beyond myocardial injury diagnosis, emphasising its value in risk stratification and the need for further large-scale studies to validate these associations and guide clinical management. Furthermore, due to limited event-timing data, some patients may have developed sepsis secondarily after admission for other acute cardiovascular conditions. As a retrospective secondary analysis, unmeasured confounding cannot be excluded despite PSM and multivariate modelling. The use of ICD-9/10 coding may also lead to diagnostic misclassification, and data on medications, ventricular function and sepsis management strategies were not fully available. Although the observed HRs (≈1.16–1.17) represent modest relative risks, they remain clinically meaningful given the multifactorial determinants of sepsis mortality. Lastly, changes in sepsis management between 2008 and 2022 may have introduced period bias. Despite these limitations, our study provides novel insight by quantifying the graded impact of multiple ASCVD burden on sepsis outcomes. The MIMIC-IV database represents data from a single tertiary care hospital in the US, which may limit the generalisability of our findings to other healthcare settings. To address this concern, noting that the single-centre and retrospective nature of the MIMIC-IV database may restrict the applicability of our results to broader or more diverse populations. We suggest that future multicentre and international studies are warranted to validate these findings across various healthcare systems and populations. The MIMIC-IV database does not provide complete data on outpatient medication use, cardiac functional parameters or metabolic control, which may have affected the precision of risk estimation. The sparseness of important clinical information in the MIMIC-IV database, such as medication use (eg, statins, antiplatelet agents, beta-blockers), ventricular function, ejection fraction, prior functional status and glycaemic control, could represent potential confounders influencing the observed associations. And the MIMIC-IV database does not contain complete details on specific sepsis management interventions, noting that the lack of information on individualised management strategies may have introduced unmeasured variability in mortality outcomes. Our study demonstrates a clear association between ASCVD and worsened sepsis outcomes, underscoring the need for enhanced risk stratification in the ICU. Given the increased mortality risk observed in patients with ASCVD, we propose that ASCVD be incorporated as a high-risk feature in sepsis management protocols to facilitate early identification and guide personalised treatment strategies. In response to our findings and in line with emerging evidence, we propose the following key areas for improvement in sepsis management as follows. (1) Risk stratification and early identification: incorporate ASCVD as a high-risk factor in ICU protocols to improve early detection and enable tailored treatments for the heightened inflammatory and thrombotic state. Enhanced screening can identify patients who may benefit from closer monitoring and aggressive management. (2) Therapeutic considerations: refine existing risk scores (eg, SOFA, SAPS II) by including ASCVD burden as an additional prognostic indicator to improve predictive accuracy for adverse outcomes. (3) Integration with sepsis guidelines: we advocate for the potential refinement of existing sepsis risk scores—such as SOFA and SAPS II—by incorporating the burden of ASCVD as an additional prognostic indicator. Including ASCVD in these scoring systems could enhance the predictive accuracy for adverse outcomes. (4) Emerging agents such as low-dose rivaroxaban (2.5 mg), SGLT2 inhibitors and GLP-1 receptor agonists have demonstrated cardiovascular protective effects beyond glycaemic control in chronic cardiovascular populations. Although their use was not analysed in the current study, these drugs may hold potential relevance for patients with concurrent sepsis and ASCVD due to their anti-inflammatory and endothelial-protective properties. However, this hypothesis requires careful validation. Future prospective studies and clinical trials are warranted to explore whether these therapies could be safely and effectively integrated into early sepsis management for this high-risk subgroup. Further research is needed to explore their potential impact on ASCVD in sepsis patients. Their inclusion helps maintain a broad perspective on the patients’ health status. Another limitation is that while the updated MIMIC-IV database includes out-of-hospital mortality data from state death records, it does not specify the cause of death. This lack of specificity limits our ability to distinguish major adverse cardiac and cerebrovascular events from other causes of death, which would have provided a more detailed understanding of the cardiovascular outcomes in sepsis patients with ASCVD. Conclusion ASCVD significantly increases mortality risk in ICU patients with sepsis, with mortality escalating alongside the number of affected vascular beds. Integrated management of sepsis and ASCVD—using targeted interventions such as tailored pharmacotherapy and risk factor modification—may reduce mortality and morbidity. Supplementary material Footnotes Funding: This work was financially supported by the College of Medicine, MacKay Medical University. (Grant no. MMC-RD-113-1E-P004 and MMC-RD-113-1C-P004). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors declare no competing interests. Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2025-108713). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: This study involves human participants and was approved by the Institutional Review Board of MacKay Memorial Hospital (IRB No. 24MMHIS460e). The requirement for informed consent was waived because the study used retrospective, de-identified clinical data. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Data availability free text: The data generated in the present study may be requested from the corresponding author. Data availability statement Data are available upon reasonable request. References Supplementary Materials Data Availability Statement Data are available upon reasonable request.