Currently, intensive care unit (ICU) patients have a higher probability of having coronary artery disease, and the mortality rate of such patients in the ICU is receiving increasing attention.

To verify whether composite inflammation indicators are significantly associated with ICU mortality in ICU patients with coronary heart disease and to develop a simple personalized prediction model

SII

SIRI

NLR

PLR

neutrophil to lymphocyte platelet ratio (NLPR)

aggregate index of systemic inflammation (AISI)

RDW

7115 patients with multi-parameter intelligent monitoring in the intensive care database IV were randomly assigned to the training cohort (n = 5692) and the internal validation cohort (n = 1423), and 701 patients in the eICU collaborative research database served as the external validation cohort

Determining the association between various inflammatory indicators and ICU mortality through multivariate logistic regression analysis and Cox proportional hazards model

A novel prediction model for mortality in ICU coronary artery disease patients was developed in the training cohort and performance evaluated in internal and external validation cohorts

To further assess the independent association between indicators and the primary endpoint, logistic regression models and Cox proportional hazards models were used, with different models used to adjust for potential confounders.

After grouping the indicators into quartiles, using G1 as the reference group, calculate the adjusted odds ratio (OR) or hazard ratio (HR) of the primary endpoint of other groups relative to the reference group.

Combining these 7 inflammation indicators with the widely used Sofa, a new ICU coronary heart disease patient's death risk prediction model was built

First, these inflammatory indicators were divided into elevated value group and non-elevated value group according to their respective 3rd quartile.

Divide the entire queue into a training queue and an internal validation queue at a ratio of 8:2

In the training cohort, we use univariate logistic regression analysis and stepwise forward multivariate logistic regression analysis to select variables for building a new prediction model, and calculate correlation coefficients and variance inflation factors (VIF) to detect the coherence of variables in the model. variance, and use the Hosmer-Lemeshow test to evaluate the fit of the logistic regression model

Logistic regression analysis and Cox proportional hazards model confirmed that various inflammatory indicators are significantly associated with ICU mortality, 30-day ICU mortality, and 90-day ICU mortality in ICU coronary heart disease patients.

The areas under the curve of the new prediction model for ICU mortality in ICU patients with coronary heart disease were 0.885 (internal validation cohort) and 0.726 (external validation cohort)

The calibration curve shows that the model's predicted probabilities match the actual observed probabilities

Decision curve analysis shows that new predictive model has high net clinical benefit

Among ICU patients with coronary heart disease, various inflammatory indicators are independent risk factors for ICU mortality.

A new ICU mortality risk prediction model for coronary heart disease patients was constructed, which has great potential to guide clinical decision-making.

Baseline characteristics of ICU patients with coronary artery disease

Correlation between various inflammatory indicators and ICU mortality in ICU patients with coronary heart disease

Model 1: Unadjusted

Model 2: Adjusting for age, gender, and race

Model 3: Adjusted for age, male gender, race, heart rate, respiratory rate, peripheral blood oxygen saturation, sequential organ failure score, acute exacerbation of chronic obstructive pulmonary disease, congestive heart failure, malignancy, dyslipidemia, acute respiratory failure, Acute renal failure, serum potassium, anion gap, blood urea nitrogen, blood glucose, serum creatinine, Hematocrit, hemoglobin, mean corpuscular volume, hematocrit, dialysis, vasopressors, invasive mechanical ventilation, coronary artery bypass grafting, angiotensin-converting enzyme inhibitors or angiotensin receptor antagonists, anti- Platelets, statins, non-vitamin K antagonist oral anticoagulants, vitamin K antagonists

Correlation between various inflammatory indicators and 30-day ICU mortality in ICU patients with coronary heart disease

Correlation between various inflammatory indicators and 90-day ICU mortality in ICU patients with coronary heart disease

Univariate and multivariate analysis of the relationship between variables and ICU mortality

These inflammatory indicators were divided into elevated value groups and non-elevated value groups according to their respective 3rd quartiles.

7 X's converted into two categories

Patient screening flow chart

Restricted cubic splines between inflammation indicators (SII, SIRI, NLR, PLR, NLPR, AISI, RDW) and ICU mortality (A), ICU 30-day mortality (B), and ICU 90-day mortality (C) function

Use RCS to explore non-linear relationships between these indicators and study outcomes when used as continuous variables

KM survival analysis of cumulative all-cause mortality in patients with coronary heart disease admitted to ICU within 90 days

The KM survival analysis method was used to analyze the cumulative rate of all-cause death in CHD patients within 90 days of admission to the ICU, and the cumulative distribution of deaths among the four score subgroups of patients with each index at admission was compared.

Correlation coefficient and variance inflation factor of each variable in the model

VIF bar chart

Nomogram of new model

ROC, calibration curve and DCA of internal validation cohort (A) and external validation cohort (B)

SOFA: Sequential Organ Failure Assessment

DCA was compared with SOFA, and the Integrated Discriminant Promotion (IDI) was calculated to verify the variability of the prediction performance of the new model between the new model and SOFA.

A user interface for a novel model-based online prediction tool