Monocyte-to-HDL-cholesterol ratio and 1-year mortality in critically ill coronary heart disease: A MIMIC cohort study Authors: Ding Li, Jiayang Dong, Zhiqiang Zhang, Xinyue Yang, Jian Zhang Published: 2026 ABSTRACT ObjectiveThe monocyte-to-high-density lipoprotein cholesterol ratio (MHR) integrates inflammatory burden with anti-inflammatory lipid status, yet its prognostic value for long-term outcomes in critically ill patients with coronary heart disease (CHD) remains unclear. This study aimed to evaluate the association between MHR and 1-year all-cause mortality in this high-risk population.MethodsWe conducted a retrospective cohort study using data from the Medical Information Mart for Intensive Care (MIMIC)-III and IV databases (2001–2019). Adult patients (≥18 years) with CHD were included, excluding those with ICU stay <24 hours or missing monocyte count/HDL-C data, yielding 3,004 patients. MHR was calculated as monocyte count (×109/L) divided by HDL-C (mmol/L). The primary outcome was 1-year all-cause mortality. Kaplan-Meier analysis and log-rank tests compared survival across MHR tertiles. Restricted cubic splines assessed dose-response relationships. Multivariable Cox proportional hazards regression with progressive adjustment evaluated MHR-mortality associations. Subgroup analyses examined effect modification across predefined clinical characteristics.ResultsThe cohort had a mean age of 73.5 years, with 65.8% males. Patients in the highest MHR tertile (T3) demonstrated greater disease severity, higher comorbidity burden, and more intensive treatment requirements. Kaplan-Meier analysis revealed significantly lower 1-year survival in T3 versus T1 (68% vs. 80%, log-rank P<0.001). Restricted cubic spline analysis showed a linear relationship (P for non-linearity=0.441). After adjustment for demographics, vital signs, comorbidities, laboratory parameters, and treatments, each 1-unit increase in MHR conferred a HR of 1.33 (95% CI: 1.18-1.51), per-SD HR = 1.10, 95% CI: 1.03-1.17. Tertile analysis demonstrated a dose-response pattern, with T3 showing HR=1.25 (95% CI: 1.04-1.51) compared with T1 (P for trend<0.05). Subgroup analyses revealed consistent associations across most strata.ConclusionsElevated MHR is associated with increased 1-year mortality in critically ill CHD patients. As a readily available composite marker, MHR may have supplementary prognostic relevance, but its clinical utility requires prospective validation. Introduction Coronary heart disease (CHD) remains the leading cause of death worldwide, accounting for approximately 9 million deaths annually and imposing substantial burdens on healthcare systems.1,2 This burden is particularly pronounced among critically ill patients requiring intensive care unit (ICU) admission, where CHD-related mortality rates exceed 20% at one year despite advances in pharmacological therapies and interventional techniques.3,4 The pathophysiology of CHD involves complex interactions between chronic inflammation, endothelial dysfunction, and lipid metabolism dysregulation.5,6 Monocytes serve as key mediators in this process, migrating into arterial walls where they differentiate into macrophages and foam cells, thereby perpetuating inflammatory cascades that promote atherosclerotic plaque formation, destabilization, and ultimately acute coronary events. 7 High-density lipoprotein cholesterol (HDL-C), in contrast, exerts atheroprotective effects through reverse cholesterol transport, antioxidant properties, and anti-inflammatory actions that counterbalance monocyte-driven pathology.8,9 This dynamic interplay between pro-inflammatory and anti-inflammatory mechanisms is particularly relevant in this high-risk population, where both heightened systemic inflammation and altered lipid profiles contribute to adverse outcomes. However, traditional single biomarkers often fail to capture this multifaceted pathophysiological interaction, suggesting that composite biomarkers integrating inflammatory burden with lipid status may offer superior prognostic discrimination in high-risk CHD patients. The monocyte-to-HDL cholesterol ratio (MHR), calculated by dividing absolute monocyte count by HDL-C concentration, has emerged as a novel and readily accessible composite inflammatory-lipid biomarker. 10 Elevated MHR has been shown to independently predict cardiovascular events and mortality across acute coronary syndromes, stable coronary artery disease, chronic kidney disease, and metabolic syndrome.11–15 A recent meta-analysis of 9 studies including 11,345 CHD patients reported pooled long-term all-cause mortality HR of 1.71 (95% CI 1.10–2.66) and long-term MACE HR of 1.72 (95% CI 1.36–2.18) for high vs. low MHR, 16 and a 2024 ACS-focused meta-analysis confirmed similar associations across in-hospital and 1-year endpoints. 17 Recent evidence has further extended MHR’s prognostic applications to additional acute cardiac outcomes, including new-onset atrial fibrillation following STEMI. 18 We note that MHR exists within a broader family of accessible inflammatory ratios—including the neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, and systemic immune-inflammation index—each with established prognostic value in critically ill populations. The distinctive conceptual feature of MHR lies in its integration of an anti-inflammatory lipid component (HDL-C) alongside an inflammatory cellular component (monocyte count), thereby capturing the balance between pro-atherogenic inflammatory drive and atheroprotective lipid defense rather than inflammatory burden alone. Despite this accumulating evidence in ambulatory and acute coronary syndrome populations, MHR’s value in this high-risk population requiring ICU admission remains inadequately characterized. The pathophysiological rationale for examining MHR in this high-risk population is reinforced by an extensive literature documenting profound functional remodeling of HDL during acute systemic inflammation. During the acute-phase response, apolipoprotein A-I is displaced from HDL by serum amyloid A, with apoA-I levels reported to decrease by approximately 73% and HDL-associated paraoxonase-1 and platelet-activating factor acetylhydrolase activities declining by 71% and 90%, respectively.19,20 The resulting acute-phase HDL particles acquire pro-inflammatory properties, with myeloperoxidase-mediated apoA-I modification, impaired cholesterol efflux capacity, and loss of antioxidant function fundamentally altering the protective biology that HDL-C concentration is conventionally presumed to represent.21–23 These mechanisms have been directly demonstrated in critically ill and septic populations, where reduced HDL-C, decreased ApoA-I, and impaired PON-1 activity correlate with higher SOFA scores, endothelial dysfunction, organ failure, and worse clinical outcomes.24–26 Consequently, in this high-risk population, HDL-C concentration alone may substantially underestimate the degree of inflammatory-lipid imbalance, and a composite marker incorporating both the inflammatory cellular driver (monocytes) and the residually available anti-inflammatory reservoir (HDL-C) may carry distinct pathophysiological meaning relative to either component measured in isolation. CHD patients requiring ICU admission represent a uniquely vulnerable population characterized by multiorgan dysfunction, hemodynamic instability, and intensive pharmacological interventions that may fundamentally alter inflammatory-lipid dynamics. In this setting, monocytes undergo heightened activation due to endotoxin exposure and hemodynamic stress, while HDL-C suffers acute functional impairment with loss of anti-inflammatory capacity. Thus, MHR may uniquely capture the convergence of atherothrombotic risk and critical illness pathophysiology. These distinctive features render extrapolation from stable outpatient cohorts potentially unreliable and highlight the need for dedicated investigation in this high-risk setting. The present study does not seek to establish the prognostic value of MHR de novo, but rather to provide the first dedicated empirical characterization of its performance in this previously uncharacterized population. Therefore, we conducted this large-scale retrospective cohort study using comprehensive data from the Medical Information Mart for Intensive Care (MIMIC)-III and IV databases. Our objectives were to evaluate the association between MHR and all-cause mortality in ICU-level CHD patients. Methods Study population This retrospective cohort study analyzed data from the Medical Information Mart for Intensive Care III and IV (MIMIC-III and MIMIC-IV) databases, containing comprehensive clinical information for patients admitted to Beth Israel Deaconess Medical Center ICUs between 2001 and 2019.27,28 All patient clinical data extracted from the MIMIC databases were fully de-identified to maintain confidentiality and strictly protect patient privacy. Database access was obtained through appropriate Collaborative Institutional Training Initiative certification (Record ID: 60071489). This study was conducted in accordance with the ethical guidelines of the Declaration of Helsinki of 1975, as revised in 2024. 29 We consecutively included adult patients (≥18 years) with coronary heart disease (CHD) diagnoses identified using International Classification of Diseases codes (ICD-9-CM and ICD-10-CM; Supplementary Table S1). Our CHD definition encompassed a broad spectrum of ischemic heart disease phenotypes, including acute myocardial infarction (I21, 410), subsequent myocardial infarction (I22), chronic ischemic heart disease (I25, 414), and angina pectoris (I20). Exclusion criteria were: (1) multiple ICU admissions (only first admission retained), (2) ICU stay <24 hours, and (3) missing monocyte count or HDL-C data. The final cohort comprised 3,004 severely ill CHD patients (Figure 1).Figure 1.The flowchart of patients’ selection. Abbreviations: CHD, coronary heart disease; ICU, intensive care unit; MIMIC, Medical Information Mart for Intensive Care; HDL-C, high-density lipoprotein cholesterol. The flowchart of patients’ selection. Abbreviations: CHD, coronary heart disease; ICU, intensive care unit; MIMIC, Medical Information Mart for Intensive Care; HDL-C, high-density lipoprotein cholesterol. Data extraction and variables Clinical data were extracted using Structured Query Language (SQL). Demographic variables included age, sex, race, and body mass index (BMI). Vital signs (mean arterial pressure, heart rate, temperature, and respiratory rate) and laboratory parameters (monocyte count, HDL-C, hemoglobin, platelet count, glucose, creatinine, and prothrombin time) were collected within 24 hours of ICU admission. When multiple measurements were available within the 24-hour window, the worst values (furthest from normal ranges) were selected to capture peak illness severity. Comorbidities were identified using ICD codes and included heart failure, hypertension, diabetes mellitus, stroke, renal failure, and chronic obstructive pulmonary disease. Treatment information encompassed angiotensin-converting enzyme inhibitors/angiotensin receptor blockers (ACEI/ARB), antiplatelet agents, β-blockers, statins, vasoactive agents (including dobutamine, dopamine, vasopressin, phenylephrine, norepinephrine, nitrates, nesiritide, and epinephrine), and reperfusion therapy (including percutaneous coronary intervention [PCI] and coronary artery bypass grafting [CABG]) (comorbidity codes in Supplementary Table S1; variable definitions in Supplementary Table S2). The Sequential Organ Failure Assessment (SOFA) score objectively assesses organ function across six domains: respiration, coagulation, liver, cardiovascular, central nervous system, and renal function. Each domain is scored 0–4 points, with higher scores indicating more severe organ dysfunction. 30 SOFA score was calculated using the worst values recorded within the first 24 hours of ICU admission, reflecting peak organ dysfunction severity during the initial stabilization window. MHR was calculated as monocyte count (×109/L) divided by HDL-C (mmol/L). Monocyte count and HDL-C were extracted as the first available measurements within the 24-hour admission window, reflecting the patient’s inflammatory-lipid status at presentation. MHR demonstrated a right-skewed distribution ranging from 0.032 to 4.558 (median 0.39, IQR 0.21-0.70; Figure S1). Patients were categorized into tertiles with cut-points at 0.21 (T1/T2) and 0.70 (T2/T3). Median MHR values were 0.16 (IQR 0.11-0.21), 0.39 (IQR 0.33-0.48), and 0.89 (IQR 0.70-1.32) for T1, T2, and T3, respectively (P<0.001). Study outcomes The primary outcome was 1-year all-cause mortality following ICU admission. All-cause mortality was selected because the MIMIC database does not provide validated cause-of-death information, and in this high-risk populations, death frequently results from multiorgan failure where cardiovascular and non-cardiovascular causes are inextricably linked. One-year mortality was ascertained through hospital records and Social Security Death Master File linkage when available. Patients were followed from the time of ICU admission until death or the end of the 1-year follow-up period. Statistical analysis Continuous variables were presented as means ± standard deviations or medians with interquartile ranges (IQRs) based on normality testing (Shapiro-Wilk test). Categorical variables were expressed as frequencies and percentages. Baseline characteristics were compared between survivors and non-survivors using the unpaired Student's t-test or Mann-Whitney U test for continuous variables and chi-square test or Fisher’s exact test for categorical variables, as appropriate. Given that a 1-unit increment in MHR encompasses the majority of the observed distribution, the MHR–mortality association was additionally expressed per standard deviation increase to facilitate clinical interpretation and enable comparison with other continuous biomarker studies. The original per 1-unit analysis was retained to maintain comparability with prior MHR literature. Patients were categorized into tertiles (T1, T2, T3) based on the distribution of MHR values. Survival analysis was conducted using the Kaplan-Meier method, with log-rank tests comparing survival curves across MHR tertiles. Dose-response relationships between MHR and 1-year mortality were assessed using restricted cubic splines. To minimize the influence of outliers on spline curve estimation and ensure stable hazard ratio estimates at the distribution tails, extreme MHR values were truncated at the 1st and 99th percentiles (winsorization). Four knots were positioned at the 5th, 35th, 65th, and 95th percentiles of the MHR distribution, with the reference point set at the median MHR value (0.392). The primary objective of the RCS analysis was to visualize the shape of the dose-response curve and to test for potential non-linearity. Non-linearity was evaluated using likelihood ratio tests comparing the spline model to a linear model. Multivariable Cox proportional hazards regression was used to analyze the association between MHR and 1-year mortality. The proportional hazards assumption was tested using Schoenfeld residuals. The global test was not statistically significant (χ2 = 2.333, df = 1, P = 0.127), and specifically, the test for MHR showed no violation of the proportional hazards assumption (χ2 = 2.333, df = 1, P = 0.127), confirming that the effect of MHR on mortality risk remained constant over the 1-year follow-up period. Progressive adjustment included: Model 1 (unadjusted); Model 2 (adjusted for demographic information: age, sex, race, and BMI); Model 3 (Model 2 plus vital signs and comorbidities: heart rate, mean arterial pressure, respiratory rate, temperature, heart failure, hypertension, diabetes, stroke, renal failure, and chronic obstructive pulmonary disease); and Model 4 (Model 3 plus laboratory tests and treatments: hemoglobin, platelet count, glucose, creatinine, prothrombin time, SOFA score, ACEI/ARB, antiplatelet agents, β-blockers, statins, vasoactive agents, and reperfusion therapy). Both continuous (per 1-unit increase) and categorical (tertile) analyses were performed. Trend tests across tertiles were conducted by treating tertile categories as continuous variables in the regression models. Subgroup analyses were performed across predefined variables including age (<65 vs. ≥65 years), sex (male vs. female), race (White vs. Others), BMI (<25 vs. ≥25 kg/m2), heart failure (yes vs. no), diabetes (yes vs. no), hypertension (yes vs. no), stroke (yes vs. no), and estimated glomerular filtration rate (eGFR; <90 vs. ≥90 mL/min/1.73 m2), database source (MIMIC-III vs. MIMIC-IV) as a proxy for treatment era, and CHD phenotypes (chronic coronary syndrome[CCS] vs. acute coronary syndrome[ACS]), with multiplicative interaction terms tested using likelihood ratio tests. The E-value, introduced by VanderWeele and Ding, represents the minimum strength of association on the risk ratio scale that an unmeasured confounder would need to have with both the exposure and the outcome, conditional on the measured covariates, to fully explain away the observed association; a higher E-value indicates greater robustness to potential unmeasured confounding. 31 Most variables had <20% missing data (Table S3). Missing data were addressed using multiple imputation with chained equations implemented through the R package mice, 32 generating 20 imputed datasets with results pooled using Rubin’s rules. As a sensitivity analysis, we repeated the primary Cox regression analyses using complete case analysis (i.e., excluding patients with missing data) to assess the robustness of our findings. All statistical analyses were performed using R software version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria). Two-sided P values <0.05 were considered statistically significant. The reporting of this retrospective cohort study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. 33 Results Baseline characteristics This study included 3,004 severely ill patients with coronary heart disease, with a mean age of 73.5 years, of whom 65.8% were male. Table 1 presents the baseline characteristics stratified by MHR tertiles (T1, T2, and T3). Compared with patients in the lowest MHR tertile (T1), those in the highest tertile (T3) exhibited greater physiological instability, characterized by significantly elevated heart rate (86.9 vs. 82.4 min-1, P < 0.001), respiratory rate (19.8 vs. 18.5 min-1, P < 0.001), and body temperature (36.7 vs. 36.5°C, P < 0.001). Moreover, the T3 group demonstrated significantly higher SOFA scores, indicating greater disease severity (P < 0.001). Notable gender differences were observed, with the T3 group having a higher proportion of males (69.9% vs. 62.6%, P = 0.002).Table 1.Baseline characteristics of the study participants.VariablesTotalT1T2T3P-value(n = 3 004)(n = 1 001)(n = 1 001)(n = 1 002) Demographic Age, (year)69.4 ± 12.369.0 ± 11.869.7 ± 12.669.4 ± 12.70.608Gender male1976 (65.8)627 (62.6)649 (64.8)700 (69.9)0.002White Race1932 (64.3)648 (64.7)667 (66.6)617 (61.6)0.058 Vital signs BMI, (kg/m2)28.2 ± 4.927.8 ± 4.828.3 ± 4.928.4 ± 4.90.020MBP, (mmHg)94.3 ± 28.594.8 ± 29.795.1 ± 28.292.9 ± 27.30.172Hear rate, (min-1)83.8 ± 17.381.3 ± 15.283.1 ± 17.186.9 ± 19.0< 0.001Temperature, (°C)36.6 ± 0.836.5 ± 0.736.5 ± 0.836.7 ± 0.8< 0.001Respiratory rate, (min-1)18.5 ± 5.617.5 ± 4.918.4 ± 5.519.8 ± 6.0< 0.001SOFA3.0 (2.0, 5.0)3.0 (2.0, 5.0)3.0 (2.0, 5.0)3.0 (2.0, 5.0)< 0.001 Comorbidities, n (%) Heart failure1551 (51.6)461 (46.1)479 (47.9)611 (61)< 0.001Diabetes1215 (40.4)354 (35.4)395 (39.5)466 (46.5)< 0.001Hypertension2366 (78.8)780 (77.9)793 (79.2)793 (79.1)0.728Renal failure687 (22.9)198 (19.8)215 (21.5)274 (27.3)< 0.001Stroke756 (25.2)202 (20.2)258 (25.8)296 (29.5)< 0.001Cancer214 (7.1)66 (6.6)66 (6.6)82 (8.2)0.279COPD737 (24.5)228 (22.8)269 (26.9)240 (24.0)0.090 Laboratory tests Hemoglobin, (g/dl)11.1 ± 2.310.9 ± 2.211.2 ± 2.211.3 ± 2.40.001Platelet, (K/μL)210.8 ± 90.9199.3 ± 87.5211.8 ± 90.3221.4 ± 93.6< 0.001Creatinine, (mg/dL)1.0 (0.8, 1.5)0.9 (0.8, 1.3)1.0 (0.8, 1.4)1.1 (0.9, 1.7)< 0.001PT, (s)15.6 ± 6.615.2 ± 4.115.3 ± 6.716.3 ± 8.3< 0.001 Treatments, n (%) ACEI/ARB799 (26.6)247 (24.7)278 (27.8)274 (27.3)0.236Anti-platelet1991 (66.3)653 (65.2)671 (67.0)667 (66.6)0.677β-blockers2132 (71.0)741 (74.0)707 (70.6)684 (68.3)0.017Statin1814 (60.4)618 (61.7)607 (60.6)589 (58.8)0.393Vasoactive agents1537 (51.2)511 (51.0)471 (47.1)555 (55.4)< 0.001Reperfusion therapy1518 (50.5)623 (62.2)520 (51.9)375 (37.4)< 0.001 PCI598 (19.9)208 (20.8)203 (20.3)187 (18.7)0.463 CABG951 (31.7)430 (43.0)327 (32.7)194 (19.4)< 0.001Note. Continuous variables are presented as mean ± standard deviation or median (interquartile range), as appropriate according to data distribution; categorical variables are presented as n (%).Abbreviations: BMI, body mass index; MBP, mean arterial pressure; SOFA, sequential organ failure assessment; COPD, chronic obstructive pulmonary disease; PT, prothrombin time; ACEI/ARB, angiotensin converting enzyme inhibitors/angiotension receptor antagonists; PCI, percutaneous coronary intervention; CABG, coronary artery bypass graft. Table 1.Baseline characteristics of the study participants.VariablesTotalT1T2T3P-value(n = 3 004)(n = 1 001)(n = 1 001)(n = 1 002) Demographic Age, (year)69.4 ± 12.369.0 ± 11.869.7 ± 12.669.4 ± 12.70.608Gender male1976 (65.8)627 (62.6)649 (64.8)700 (69.9)0.002White Race1932 (64.3)648 (64.7)667 (66.6)617 (61.6)0.058 Vital signs BMI, (kg/m2)28.2 ± 4.927.8 ± 4.828.3 ± 4.928.4 ± 4.90.020MBP, (mmHg)94.3 ± 28.594.8 ± 29.795.1 ± 28.292.9 ± 27.30.172Hear rate, (min-1)83.8 ± 17.381.3 ± 15.283.1 ± 17.186.9 ± 19.0< 0.001Temperature, (°C)36.6 ± 0.836.5 ± 0.736.5 ± 0.836.7 ± 0.8< 0.001Respiratory rate, (min-1)18.5 ± 5.617.5 ± 4.918.4 ± 5.519.8 ± 6.0< 0.001SOFA3.0 (2.0, 5.0)3.0 (2.0, 5.0)3.0 (2.0, 5.0)3.0 (2.0, 5.0)< 0.001 Comorbidities, n (%) Heart failure1551 (51.6)461 (46.1)479 (47.9)611 (61)< 0.001Diabetes1215 (40.4)354 (35.4)395 (39.5)466 (46.5)< 0.001Hypertension2366 (78.8)780 (77.9)793 (79.2)793 (79.1)0.728Renal failure687 (22.9)198 (19.8)215 (21.5)274 (27.3)< 0.001Stroke756 (25.2)202 (20.2)258 (25.8)296 (29.5)< 0.001Cancer214 (7.1)66 (6.6)66 (6.6)82 (8.2)0.279COPD737 (24.5)228 (22.8)269 (26.9)240 (24.0)0.090 Laboratory tests Hemoglobin, (g/dl)11.1 ± 2.310.9 ± 2.211.2 ± 2.211.3 ± 2.40.001Platelet, (K/μL)210.8 ± 90.9199.3 ± 87.5211.8 ± 90.3221.4 ± 93.6< 0.001Creatinine, (mg/dL)1.0 (0.8, 1.5)0.9 (0.8, 1.3)1.0 (0.8, 1.4)1.1 (0.9, 1.7)< 0.001PT, (s)15.6 ± 6.615.2 ± 4.115.3 ± 6.716.3 ± 8.3< 0.001 Treatments, n (%) ACEI/ARB799 (26.6)247 (24.7)278 (27.8)274 (27.3)0.236Anti-platelet1991 (66.3)653 (65.2)671 (67.0)667 (66.6)0.677β-blockers2132 (71.0)741 (74.0)707 (70.6)684 (68.3)0.017Statin1814 (60.4)618 (61.7)607 (60.6)589 (58.8)0.393Vasoactive agents1537 (51.2)511 (51.0)471 (47.1)555 (55.4)< 0.001Reperfusion therapy1518 (50.5)623 (62.2)520 (51.9)375 (37.4)< 0.001 PCI598 (19.9)208 (20.8)203 (20.3)187 (18.7)0.463 CABG951 (31.7)430 (43.0)327 (32.7)194 (19.4)< 0.001Note. Continuous variables are presented as mean ± standard deviation or median (interquartile range), as appropriate according to data distribution; categorical variables are presented as n (%).Abbreviations: BMI, body mass index; MBP, mean arterial pressure; SOFA, sequential organ failure assessment; COPD, chronic obstructive pulmonary disease; PT, prothrombin time; ACEI/ARB, angiotensin converting enzyme inhibitors/angiotension receptor antagonists; PCI, percutaneous coronary intervention; CABG, coronary artery bypass graft. Baseline characteristics of the study participants. Note. Continuous variables are presented as mean ± standard deviation or median (interquartile range), as appropriate according to data distribution; categorical variables are presented as n (%). Abbreviations: BMI, body mass index; MBP, mean arterial pressure; SOFA, sequential organ failure assessment; COPD, chronic obstructive pulmonary disease; PT, prothrombin time; ACEI/ARB, angiotensin converting enzyme inhibitors/angiotension receptor antagonists; PCI, percutaneous coronary intervention; CABG, coronary artery bypass graft. Patients in the highest MHR tertile demonstrated a substantially greater burden of cardiovascular and systemic comorbidities. The prevalence of heart failure increased progressively from T1 to T3 (46.1% vs. 61.0%, P < 0.001), as did diabetes (35.4% vs. 46.5%, P < 0.001), renal failure (19.8% vs. 27.3%, P < 0.001), and stroke (9.8% vs. 14.8%, P < 0.001). Laboratory parameters further reflected this heightened disease burden, with patients in T3 presenting with elevated hemoglobin, platelet count, and creatinine levels, as well as prolonged prothrombin time (all P ≤ 0.001), indicating more severe hematological abnormalities and coagulation dysfunction. Treatment patterns revealed substantial disparities across MHR tertiles. Patients in the T3 group received significantly less coronary revascularization, with CABG rates markedly lower compared to the T1 group (19.4% vs. 43.0%, P < 0.001). Conversely, they required more intensive hemodynamic support, as evidenced by significantly higher vasoactive agent administration (55.4% vs. 43.1%, P < 0.001). Additionally, β-blocker utilization declined from T1 to T3 (74.0% vs. 68.3%, P = 0.017), suggesting potential contraindications or greater clinical complexity in patients with higher MHR. Survival analysis and dose-response relationships Kaplan-Meier survival analysis demonstrated a significant association between MHR tertiles and 1-year mortality in this high-risk population (Figure 2). Patients in the highest MHR tertile (T3) exhibited markedly lower survival rates compared with those in the middle (T2) and lowest (T1) tertiles during the 1-year follow-up period (log-rank P < 0.001). The survival curves diverged early after study entry and remained persistently separated throughout the observation period. At 360 days, the survival probability was approximately 68% in the T3 group, versus 76% in the T2 group and 80% in the T1 group, demonstrating a graded association between higher MHR levels and increased mortality risk. The number of patients at risk declined more rapidly in the T3 group, with 677, 764, and 804 patients remaining at 360 days in the T3, T2, and T1 groups, respectively.Figure 2.Kaplan-Meier survival curves for 1-year mortality stratified by MHR tertiles. The figure displays survival probability over 360 days following study entry for critically ill patients with coronary heart disease, grouped by MHR tertiles (T1, lowest tertile; T2, middle tertile; T3, highest tertile). The green line represents T1, the blue line represents T2, and the red line represents T3. The inset shows a magnified view of the later follow-up period to facilitate visualization of survival curve separation across MHR tertiles. The number of patients at risk at each time point is shown below the x-axis. Statistical significance was assessed using the log-rank test (P < 0.001). Abbreviations: MHR, monocyte to high-density lipoprotein cholesterol ratio. Kaplan-Meier survival curves for 1-year mortality stratified by MHR tertiles. The figure displays survival probability over 360 days following study entry for critically ill patients with coronary heart disease, grouped by MHR tertiles (T1, lowest tertile; T2, middle tertile; T3, highest tertile). The green line represents T1, the blue line represents T2, and the red line represents T3. The inset shows a magnified view of the later follow-up period to facilitate visualization of survival curve separation across MHR tertiles. The number of patients at risk at each time point is shown below the x-axis. Statistical significance was assessed using the log-rank test (P < 0.001). Abbreviations: MHR, monocyte to high-density lipoprotein cholesterol ratio. Dose-response relationship between MHR and 1-year mortality To investigate the presence of a dose-response relationship between MHR and 1-year mortality, we conducted a restricted cubic spline analysis with the reference point set at 0.392. Following adjustment for potential confounders, the relationship between MHR and mortality appeared linear, with no evidence of non-linearity detected (P for non-linearity = 0.441, Figure 3).Figure 3.Dose-response relationship between MHR and 1-year all-cause mortality. The figure illustrates the adjusted hazard ratio (solid red line) with 95% confidence intervals (shaded pink area) for the association between MHR and 1-year mortality using restricted cubic spline regression with four knots. The reference point was set at the median MHR value (0.392, bold vertical dashed line). The horizontal line at HR = 1.0 is shown as the primary reference line, whereas the vertical dashed lines indicate the prespecified knot locations. The blue histogram at the bottom displays the distribution of MHR values among study participants. Four knots were positioned at the 5th, 35th, 65th, and 95th percentiles of the MHR distribution (thin vertical dashed lines with corresponding MHR values labeled). The test for non-linearity was not significant (P = 0.441), confirming a linear relationship. The model was adjusted for covariates including demographics, vital signs, comorbidities, laboratory tests, and treatments. Abbreviations: MHR, monocyte to high-density lipoprotein cholesterol ratio. Dose-response relationship between MHR and 1-year all-cause mortality. The figure illustrates the adjusted hazard ratio (solid red line) with 95% confidence intervals (shaded pink area) for the association between MHR and 1-year mortality using restricted cubic spline regression with four knots. The reference point was set at the median MHR value (0.392, bold vertical dashed line). The horizontal line at HR = 1.0 is shown as the primary reference line, whereas the vertical dashed lines indicate the prespecified knot locations. The blue histogram at the bottom displays the distribution of MHR values among study participants. Four knots were positioned at the 5th, 35th, 65th, and 95th percentiles of the MHR distribution (thin vertical dashed lines with corresponding MHR values labeled). The test for non-linearity was not significant (P = 0.441), confirming a linear relationship. The model was adjusted for covariates including demographics, vital signs, comorbidities, laboratory tests, and treatments. Abbreviations: MHR, monocyte to high-density lipoprotein cholesterol ratio. Association between MHR and 1-year mortality Progressive multivariable Cox regression analysis demonstrated a significant association between MHR and 1-year mortality (Table 2). Each 1-uint increase in MHR was associated with a fully adjusted hazard ratio of 1.33 (95% CI: 1.18–1.51), indicating a 33% increase in 1-year mortality risk per 1-uint increment. Each 1 SD increase in MHR was associated with a fully adjusted hazard ratio of 1.10 (95% CI: 1.03–1.17, P<0.001), indicating a 10% increase in 1-year mortality risk per SD increment. Tertile analysis revealed a dose-response pattern: compared with T1 (reference), T3 showed an adjusted HR of 1.25 (95% CI: 1.04-1.51), whereas T2 demonstrated no significant association (HR = 1.10, 95% CI: 0.91-1.33). The trend test confirmed a significant graded relationship across tertiles (HR per tertile increase = 1.12, 95% CI: 1.02-1.23, P for trend < 0.05), indicating that elevated MHR was independently associated with increased 1-year mortality risk in this high-risk population.Table 2.Association of MHR with 1-year all-cause mortality.MHRn/N (%)Model 1Model 2Model 3Model 4Per-unit increase762/3004 (25.4)1.48 (1.34, 1.63)1.56 (1.41, 1.72)1.32 (1.18, 1.47)1.33 (1.18, 1.51)Per-SD increase1.24 (1.18, 1.32)1.29 (1.22, 1.36)1.17 (1.10, 1.24)1.10 (1.03, 1.17)T1197/1001 (19.7)Ref. (1.00)Ref. (1.00)Ref. (1.00)Ref. (1.00)T2197/1001 (23.9)1.24 (1.03, 1.50)1.28 (1.06, 1.54)1.14 (0.95, 1.39)1.10 (0.91, 1.33)T3326/1002 (32.5)1.80 (1.51, 2.15)1.90 (1.59, 2.27)1.42 (1.18, 1.71)1.25 (1.04, 1.51)Trend. test​1.35 (1.24, 1.48)1.39 (1.27, 1.52)1.19 (1.09, 1.31)1.12 (1.02, 1.23)Note. Model 1 adjust for: none; Model 2 adjusted for: demographic information (age, sex, race, and BMI); Model 3 adjusted for: Model 2 + vital signs (mean arterial pressure, heart rate, temperature, respiratory rate, and SOFA score) + comorbidities (heart failure, diabetes, hypertension, renal failure, stroke, cancer, and COPD) + laboratory tests (hemoglobin, platelet, creatinine, and prothrombin time); Model 4 adjusted for: Model 3 + treatments (ACEI/ARB, antiplatelet agents, β-blockers, statins, vasoactive agents, and reperfusion therapy).Abbreviations: MHR, monocyte to high-density lipoprotein cholesterol ratio. HR, hazard ratio; CI, confidence interval. Table 2.Association of MHR with 1-year all-cause mortality.MHRn/N (%)Model 1Model 2Model 3Model 4Per-unit increase762/3004 (25.4)1.48 (1.34, 1.63)1.56 (1.41, 1.72)1.32 (1.18, 1.47)1.33 (1.18, 1.51)Per-SD increase1.24 (1.18, 1.32)1.29 (1.22, 1.36)1.17 (1.10, 1.24)1.10 (1.03, 1.17)T1197/1001 (19.7)Ref. (1.00)Ref. (1.00)Ref. (1.00)Ref. (1.00)T2197/1001 (23.9)1.24 (1.03, 1.50)1.28 (1.06, 1.54)1.14 (0.95, 1.39)1.10 (0.91, 1.33)T3326/1002 (32.5)1.80 (1.51, 2.15)1.90 (1.59, 2.27)1.42 (1.18, 1.71)1.25 (1.04, 1.51)Trend. test​1.35 (1.24, 1.48)1.39 (1.27, 1.52)1.19 (1.09, 1.31)1.12 (1.02, 1.23)Note. Model 1 adjust for: none; Model 2 adjusted for: demographic information (age, sex, race, and BMI); Model 3 adjusted for: Model 2 + vital signs (mean arterial pressure, heart rate, temperature, respiratory rate, and SOFA score) + comorbidities (heart failure, diabetes, hypertension, renal failure, stroke, cancer, and COPD) + laboratory tests (hemoglobin, platelet, creatinine, and prothrombin time); Model 4 adjusted for: Model 3 + treatments (ACEI/ARB, antiplatelet agents, β-blockers, statins, vasoactive agents, and reperfusion therapy).Abbreviations: MHR, monocyte to high-density lipoprotein cholesterol ratio. HR, hazard ratio; CI, confidence interval. Association of MHR with 1-year all-cause mortality. Note. Model 1 adjust for: none; Model 2 adjusted for: demographic information (age, sex, race, and BMI); Model 3 adjusted for: Model 2 + vital signs (mean arterial pressure, heart rate, temperature, respiratory rate, and SOFA score) + comorbidities (heart failure, diabetes, hypertension, renal failure, stroke, cancer, and COPD) + laboratory tests (hemoglobin, platelet, creatinine, and prothrombin time); Model 4 adjusted for: Model 3 + treatments (ACEI/ARB, antiplatelet agents, β-blockers, statins, vasoactive agents, and reperfusion therapy). Abbreviations: MHR, monocyte to high-density lipoprotein cholesterol ratio. HR, hazard ratio; CI, confidence interval. Subgroup analysis Subgroup analyses evaluated the consistency of the MHR-mortality association across predefined patient populations (Figure 4, and Figure S3). MHR maintained consistent prognostic value across age (<65 vs. ≥65 years), race (White vs. Others), BMI (<25 vs. ≥25 kg/m2), heart failure, diabetes, hypertension, stroke, eGFR (<90 vs. ≥90 mL/min/1.73 m2), database source (MIMIC-III vs. MIMIC-IV), and CHD phenotypes (CCS vs. ACS), subgroups (all P values for interaction >0.05). An exploratory sex interaction was observed (P for interaction = 0.001), with a stronger association observed in female patients (HR=1.34, 95% CI: 1.21–1.48) compared with male patients (HR=1.00, 95% CI: 0.93–1.09), ratio of HRs: 1.34 (95% CI: 1.18–1.52).Figure 4.Subgroup analysis of 1-year mortality. Forest plot showing hazard ratios with 95% confidence intervals for the association between MHR (per 1-unit increase) and 1-year mortality across predefined subgroups. P values for interaction assess effect modification. Models were adjusted for demographics, vital signs, comorbidities, laboratory tests, and treatments. Abbreviations: BMI, body mass index; eGFR, glomerular filtration rate; HR, hazard ratio; CI, confidence interval. Subgroup analysis of 1-year mortality. Forest plot showing hazard ratios with 95% confidence intervals for the association between MHR (per 1-unit increase) and 1-year mortality across predefined subgroups. P values for interaction assess effect modification. Models were adjusted for demographics, vital signs, comorbidities, laboratory tests, and treatments. Abbreviations: BMI, body mass index; eGFR, glomerular filtration rate; HR, hazard ratio; CI, confidence interval. E-value analysis To assess the robustness of our findings against potential unmeasured confounding, we performed E-value analysis (Figure S2). The E-value for the point estimate was 1.73, and the E-value for the lower bound of the 95% confidence interval was 1.49. Sensitivity analysis Sensitivity analysis using complete case analysis (n = 2,077) yielded consistent results with the primary analysis. After excluding patients with missing data, each 1- SD increase in MHR remained significantly associated with increased 1-year mortality risk (HR = 1.15, 95% CI: 1.07–1.24) in the fully adjusted model. The dose-response pattern across tertiles was also confirmed (T3 vs. T1: HR = 1.41, 95% CI: 1.12–1.78; P for trend < 0.05), supporting the robustness of our primary findings (Table 3).Table 3.Association of MHR with 1-year all-cause mortality after dropping missing data.MHRn/N (%)Model 1Model 4Per-unit increase518/2077 (24.9)1.59 (1.41∼1.78)1.28 (1.12∼1.47)Per-SD increase1.30 (1.22∼1.39)1.15 (1.07∼1.24)T1121/692 (17.5)Ref. (1.00)Ref. (1.00)T2167/692 (24.1)1.45 (1.15∼1.83)1.21 (0.96∼1.54)T3230/693 (33.2)2.10 (1.69∼2.62)1.41 (1.12∼1.78)Trend. test​1.45 (1.30∼1.62)1.19 (1.06∼1.33)Note. Model 1 adjust for: none; Model 4 adjusted for covariates included in demographics, vital signs, comorbidities, laboratory tests and treatments.Abbreviations: MHR, monocyte to high-density lipoprotein cholesterol ratio; OR, odds ratio; CI, confidence interval; HR, hazard ratio. Table 3.Association of MHR with 1-year all-cause mortality after dropping missing data.MHRn/N (%)Model 1Model 4Per-unit increase518/2077 (24.9)1.59 (1.41∼1.78)1.28 (1.12∼1.47)Per-SD increase1.30 (1.22∼1.39)1.15 (1.07∼1.24)T1121/692 (17.5)Ref. (1.00)Ref. (1.00)T2167/692 (24.1)1.45 (1.15∼1.83)1.21 (0.96∼1.54)T3230/693 (33.2)2.10 (1.69∼2.62)1.41 (1.12∼1.78)Trend. test​1.45 (1.30∼1.62)1.19 (1.06∼1.33)Note. Model 1 adjust for: none; Model 4 adjusted for covariates included in demographics, vital signs, comorbidities, laboratory tests and treatments.Abbreviations: MHR, monocyte to high-density lipoprotein cholesterol ratio; OR, odds ratio; CI, confidence interval; HR, hazard ratio. Association of MHR with 1-year all-cause mortality after dropping missing data. Note. Model 1 adjust for: none; Model 4 adjusted for covariates included in demographics, vital signs, comorbidities, laboratory tests and treatments. Abbreviations: MHR, monocyte to high-density lipoprotein cholesterol ratio; OR, odds ratio; CI, confidence interval; HR, hazard ratio. Discussion In this retrospective cohort study of 3,004 critically ill CHD patients from the MIMIC-III and IV databases, we demonstrated that elevated MHR independently predicted 1-year mortality. After comprehensive adjustment for demographic, clinical, and therapeutic factors, each 1-unit MHR increase was associated with a 33% higher mortality risk (HR=1.33, 95% CI: 1.18-1.51, P<0.001). This association was corroborated by Kaplan-Meier analysis, which revealed significantly lower survival in the highest versus lowest MHR tertile (68% vs. 80%, log-rank P<0.001). Restricted cubic spline analysis confirmed a linear dose-response relationship without evidence of threshold effects (P for non-linearity=0.441). Most importantly, subgroup analyses uncovered significant sex-based effect modification (P for interaction=0.001), with substantially stronger associations observed in females (HR=1.34, 95% CI: 1.21–1.48) compared to males (HR=1.00, 95% CI: 0.93–1.09), while associations remained consistent across other clinical subgroups. Collectively, these findings establish MHR as a candidate prognostic marker in this high-risk patients. Our findings align with previous investigations demonstrating the prognostic utility of MHR in cardiovascular populations, while also revealing several critical distinctions. Kanbay et al. first reported that elevated MHR independently predicted major adverse cardiovascular events in patients with acute coronary syndrome, with hazard ratios ranging from 2.24 to 4.91 depending on the outcome assessed. 10 Similarly, Cetin et al. demonstrated that MHR was associated with cardiovascular mortality in stable coronary artery disease patients, 12 while Açıkgöz et al. observed comparable associations in patients undergoing percutaneous coronary intervention (HR=2.0, 95% CI: 1.2-4.1). 34 Our observed effect size is more modest, likely reflecting the multifactorial nature of mortality in critically ill populations, where competing risks from multiorgan dysfunction and hemodynamic instability may dilute the specific cardiovascular signal of MHR. Our study population differs fundamentally from prior investigations, which predominantly enrolled ambulatory or acute coronary syndrome patients, in that we specifically targeted critically ill CHD patients requiring ICU admission—a substantially higher-risk phenotype with baseline mortality rates exceeding 20%. As part of our prespecified subgroup analyses, an exploratory examination of sex-specific effects was conducted. The observed interaction—with numerically stronger associations in females (HR=1.34, 95% CI: 1.21–1.48) than males (HR=1.00, 95% CI: 0.93–1.09)—is strictly hypothesis-generating and was not a prespecified primary hypothesis. An exploratory sex interaction was observed (P for interaction = 0.001), which survived Bonferroni correction for nine comparisons (adjusted α = 0.0056). This finding was not prespecified, arose from one of nine subgroup comparisons, and lacks empirical support from hormonal or sex-specific treatment data in our cohort. It should be treated as a hypothesis-generating observation requiring independent replication in prospective studies specifically designed to test sex-stratified effects. Several methodological approaches employed in our study enhance the robustness and clinical interpretability of these findings. First, we utilized comprehensive data from MIMIC-III and IV databases spanning nearly two decades, providing sufficient statistical power while implementing progressive adjustment across four hierarchical models that controlled for demographics, vital signs, comorbidities, laboratory parameters, disease severity, and treatments. This rigorous adjustment strategy ensures that the observed MHR-mortality association is independent of established prognostic factors in critically ill populations. Second, we employed multiple complementary analytical approaches—including tertile analysis, continuous variable modeling, and restricted cubic splines—which collectively demonstrated a linear dose-response relationship without threshold effects (P for non-linearity=0.441). This linearity facilitates straightforward clinical interpretation and suggests that MHR’s prognostic value extends across the entire distribution of values. Third, our comprehensive baseline characterization provided important pathophysiological context for understanding elevated MHR. Patients in the highest MHR tertile exhibited not only greater disease severity reflected by elevated SOFA scores and higher comorbidity burden, but also markedly different treatment patterns, including significantly lower coronary revascularization rates (19.4% vs. 43.0% for CABG) alongside higher vasoactive support requirements (55.4% vs. 43.1%). These patterns suggest that elevated MHR may function as an integrative marker reflecting both underlying disease severity and limited therapeutic opportunities, which may explain its strong independent prognostic value in severely ill populations. These findings suggest that MHR captures information along two interrelated axes: the inflammatory-lipid pathophysiology central to CHD, and the overall burden of disease severity and frailty. Observational data cannot disentangle these axes nor determine which predominates. However, this ambiguity does not diminish MHR’s clinical relevance—a prognostic marker’s value lies in reliably identifying high-risk patients, not in resolving its causal architecture. From a clinical perspective, all findings from this retrospective observational study are strictly hypothesis-generating. MHR’s potential value lies in its ability to identify patients at elevated mortality risk using readily available laboratory data, thereby complementing—not replacing—established ICU severity assessments such as the SOFA score. The addition of MHR to a comprehensive clinical model yielded a C-statistic increment of 0.010 and statistically significant NRI and IDI values, supporting a modest supplementary prognostic role. Whether MHR-based risk stratification can inform targeted clinical interventions, and at what thresholds, remains entirely unknown and requires investigation through prospective studies with prespecified intervention protocols and formal threshold validation before any clinical application can be considered. Limitations and future directions Several limitations warrant acknowledgment. First, the retrospective observational design precludes causal inference. Despite adjustment for SOFA score, vasoactive agent requirements, and revascularization therapy, residual confounding from unmeasured dimensions of illness severity—including frailty, functional reserve, and nutritional status—cannot be excluded, as MHR may partly reflect overall critical illness burden rather than a specific inflammatory-lipid pathway. Elevated MHR may represent an independent biological risk marker, a surrogate of illness severity, or both—and observational data alone cannot definitively distinguish among these possibilities. Second, our data derive from a single U.S. tertiary care center with limited ethnic diversity, potentially limiting generalizability to other populations and healthcare settings with different practice patterns. Specifically, academic tertiary centers tend to have higher revascularization rates, more intensive monitoring, and greater subspecialty access than community hospitals, factors that may influence MHR distributions and treatment-related confounding; additionally, the predominantly elderly cohort (mean age 69.4 years) limits extrapolation to younger CHD populations with distinct comorbidity and inflammatory-lipid profiles. Third, despite comprehensive multivariable adjustment, residual confounding from unmeasured variables—including specific inflammatory biomarkers, medication adherence, and socioeconomic factors—cannot be entirely excluded. The E-value for the point estimate was 1.73, and the E-value for the lower bound of the 95% confidence interval was 1.49. However, this magnitude is modest and falls within the plausible range of unmeasured confounders such as frailty, nutritional status, or functional capacity. The E-value quantifies the minimum confounding strength required to explain away the observed association, which in this case is not large; therefore, unmeasured confounding remains a plausible alternative explanation that cannot be excluded. Fourth, MHR was measured within 24 hours of ICU admission, capturing baseline inflammatory-lipid status at presentation rather than dynamic ICU-acquired changes. This timing may be clinically meaningful, as admission MHR likely reflects the cumulative burden of chronic disease and acute decompensation severity. However, serial measurements during hospitalization and post-discharge could provide incremental prognostic information and should be explored in future studies. Fifth, our study period (2001-2019) encompasses substantial evolution in CHD management strategies, potentially introducing temporal heterogeneity in treatment effects. Subgroup analysis stratified by database source (MIMIC-III vs. MIMIC-IV) revealed no significant interaction (P for interaction=0.291), and our fully adjusted model incorporated key evolving treatment variables, providing implicit control for era-related variation. However, this subgroup comparison is a crude temporal proxy, limited by smaller MIMIC-III sample size and imperfect era separation; notably, the directionally discordant MIMIC-III point estimate (HR=0.95) suggests MHR’s prognostic signal may be stronger in the contemporary era, warranting caution in interpreting temporal consistency. Nevertheless, residual temporal confounding cannot be entirely excluded. Finally, while we identified significant sex-based effect modification, the underlying biological mechanisms—including potential contributions from hormonal modulation of HDL function, sex differences in monocyte phenotypes, or treatment disparities—remain incompletely understood and require dedicated mechanistic investigation. These limitations point toward five prioritized directions for future research. First, external validation in independent multicenter ICU cohorts across diverse geographic regions and healthcare systems is essential. Second, mechanistic investigations should prioritize elucidating the biological pathways underlying the exploratory sex-specific associations, including differences in HDL particle functionality, monocyte phenotypes, and hormonal influences on lipid metabolism. Third, studies examining dynamic MHR trajectories during ICU stay may reveal whether temporal trends provide incremental prognostic information beyond baseline values. Fourth, head-to-head comparative studies evaluating MHR against alternative inflammatory biomarkers (NLR, CRP, procalcitonin) and established ICU severity scores (APACHE II/III/IV) in prospective cohorts with complete biomarker ascertainment would clarify MHR’s relative and incremental position within the broader prognostic toolkit. Finally, interventional studies evaluating whether MHR-guided therapeutic strategies can improve outcomes in high-risk patients would establish clinical utility and inform evidence-based practice. Conclusions Elevated MHR may be associated with increased 1-year mortality in critically ill CHD patients after comprehensive multivariable adjustment, showing a potential dose-response relationship. As a readily available composite biomarker derived from routine laboratory tests, MHR may hold supplementary prognostic relevance for risk stratification in intensive care settings, though its clinical utility requires further validation in prospective studies. Supplemental material Supplemental material - Monocyte-to-HDL-cholesterol ratio and 1-year mortality in critically Ill coronary heart disease: A MIMIC cohort study Supplemental material for Monocyte-to-HDL-cholesterol ratio and 1-year mortality in critically Ill coronary heart disease: A MIMIC cohort study by Ding Li, Jiayang Dong, Zhiqiang Zhang, Xinyue Yang, and Jian Zhang in Science Progress.