Information Mart For Intensive Care Research Articles

Relationship between time-weighted average glucose and mortality in critically ill patients: a retrospective analysis of the MIMIC-IV database

Blood glucose management in intensive care units (ICU) remains a controversial topic. We assessed the association between time-weighted average glucose (TWAG) levels and ICU mortality in critically ill patients in a real-world study. This retrospective study included critically ill patients from the Medical Information Mart for Intensive Care IV database. Glycemic distance is the difference between TWAG in the ICU and preadmission usual glycemia assessed with glycated hemoglobin at ICU admission. The TWAG and glycemic distance were divided into 4 groups and 3 groups, and their associations with ICU mortality risk were evaluated using multivariate logistic regression. Restricted cubic splines were used to explore the non-linear relationship. A total of 4737 adult patients were included. After adjusting for covariates, compared with TWAG ≤ 110 mg/dL, the odds ratios (ORs) of the TWAG > 110 mg/dL groups were 1.62 (95% CI 0.97–2.84, p = 0.075), 3.41 (95% CI 1.97–6.15, p < 0.05), and 6.62 (95% CI 3.6–12.6, p < 0.05). Compared with glycemic distance at − 15.1–20.1 mg/dL, the ORs of lower or higher groups were 0.78 (95% CI 0.50–1.21, p = 0.3) and 2.84 (95% CI 2.12–3.82, p < 0.05). The effect of hyperglycemia on ICU mortality was more pronounced in non-diabetic and non-septic patients. TWAG showed a U-shaped relationship with ICU mortality risk, and the mortality risk was minimal at 111 mg/dL. Maintaining glycemic distance ≤ 20.1 mg/dL may be beneficial. In different subgroups, the impact of hyperglycemia varied.

Effect of early serum phosphate disorder on in-hospital and 28-day mortality in sepsis patients: a retrospective study based on MIMIC-IV database

BackgroundThis study aims to assess the influence of early serum phosphate fluctuation on the short-term prognosis of sepsis patients.MethodsThis retrospective study used the Medical Information Mart for Intensive Care IV database to analyze serum phosphate levels in sepsis patients within 3 days of ICU admission. According to the absolute value of delta serum phosphate (the maximum value minus the minimum value of serum phosphorus measured within three days), the patients were divided into four groups, 0–1.3, 1.4–2.0, 2.1–3.1, and ≥ 3.2 mg/dl. Meanwhile, the direction of delta serum phosphate was compared. With the serum phosphate change group of 0–1.3 mg/dl as the reference group, the relationship between delta serum phosphate and in-hospital mortality and 28-day mortality was analyzed by multivariate Logistics regression analysis.ResultsThe study involved 1375 sepsis patients. Serum phosphate changes (0–1.3, 1.4–2.0, 2.1–3.1, and ≥ 3.2 mg/dl) correlated with in-hospital and 28-day mortality variations (p = 0.005, p = 0.008). Much higher serum phosphate fluctuation elevated in-hospital and 28-day mortality. Compared to the 0–1.3 mg/dl change group, adjusted odds ratios (OR) in other groups for in-hospital mortality were 1.25 (0.86–1.81), 1.28 (0.88–1.86), and 1.63 (1.10–2.43), and for 28-day mortality were 1.21 (0.86–1.72), 1.10 (0.77–1.57), and 1.49 (1.03–2.19). Under the trend of increasing serum phosphate, the ORs of in-hospital mortality and 28-day mortality in ≥ 3.2 mg/dl group were 2.52 and 2.01, respectively.ConclusionIn conclude, the delta serum phosphate ≥ 3.2 mg/dl was associated with in-hospital mortality and 28-day mortality in patients with sepsis.

The Prognostic Value of Red Cell Distribution Width in Critically Ill Cerebral Infarction Patients: A Retrospective Cohort Study

Objective: Red blood cell distribution width (RDW) has been utilized as a prognostic indicator for mortality risk assessment in cardiovascular and cerebrovascular patients. Nevertheless, the prognostic significance of RDW in critically ill patients with cerebral infarction is yet to be investigated. The objective of this study is to examine the association between RDW and the risk of all-cause mortality in cerebral infarction patients admitted to the intensive care unit (ICU). Method: A retrospective cohort study was conducted using the Medical Information Mart for Intensive Care IV 2.2 (MIMIC-IV) intensive care dataset for data analysis. The main results were the all-cause mortality rates at 3 and 12 months of follow-up. Cumulative curves were plotted using the Kaplan-Meier method, and Cox proportional hazards analysis was used to examine the relationship between RDW and mortality rates in critically ill cerebral infarction patients. Results: The findings indicate that RDW serves as a significant prognostic factor for mortality risk in critically ill stroke patients, specifically at the 3 and 12-month follow-up periods. The observed correlation between increasing RDW levels and higher mortality rates among cerebral infarction patients further supports the potential utility of RDW as a predictive indicator. Conclusion: RDW emerges as an independent predictor of mortality risk during the 3 and 12-month follow-up periods for critically ill patients with cerebral infarction.

An early warning model for predicting major adverse kidney events within 30 days in sepsis patients.

In sepsis patients, kidney damage is among the most dangerous complications, with a high mortality rate. In addition, major adverse kidney events within 30 days (MAKE30) served as a comprehensive and unbiased clinical outcome measure for sepsis patients due to the recent shift toward targeting patient-centered renal outcomes in clinical research. However, the underlying predictive model for the prediction of MAKE30 in sepsis patients has not been reported in any study. A cohort of 2,849 sepsis patients from the Medical Information Mart for Intensive Care (MIMIC)-IV database was selected and subsequently allocated into a training set (n = 2,137, 75%) and a validation set (n = 712, 25%) through randomization. In addition, 142 sepsis patients from the Xi'An No. 3 Hospital as an external validation group. Univariate and multivariate logistic regression analyses were conducted to ascertain the independent predictors of MAKE30. Subsequently, a nomogram was developed utilizing these predictors, with an area under curve (AUC) above 0.6. The performance of nomogram was assessed through calibration curve, receiver operating characteristics (ROC) curve, and decision curve analysis (DCA). The secondary outcome was 30-day mortality, persistent renal dysfunction (PRD), and new renal replacement therapy (RRT). MAKE30 were a composite of death, PRD, new RRT. The construction of the nomogram was based on several independent predictors (AUC above 0.6), including age, respiratory rate (RR), PaO2, lactate, and blood urea nitrogen (BUN). The predictive model demonstrated satisfactory discrimination for MAKE30, with an AUC of 0.740, 0.753, and 0.821 in the training, internal validation, and external validation cohorts, respectively. Furthermore, the simple prediction model exhibited superior predictive value compared to the SOFA model in both the training (AUC = 0.710) and validation (AUC = 0.692) cohorts. The nomogram demonstrated satisfactory calibration and clinical utility as evidenced by the calibration curve and DCA. Additionally, the predictive model exhibited excellent accuracy in forecasting 30-day mortality (AUC = 0.737), PRD (AUC = 0.639), and new RRT (AUC = 0.846) within the training dataset. Additionally, the model displayed predictive power for 30-day mortality (AUC = 0.765), PRD (AUC = 0.667), and new RRT (AUC = 0.783) in the validation set. The proposed nomogram holds the potential to estimate the risk of MAKE30 promptly and efficiently in sepsis patients within the initial 24 h of admission, thereby equipping healthcare professionals with valuable insights to facilitate personalized interventions.

Stress hyperglycemia is associated with poor outcome in critically ill patients with pulmonary hypertension.

Stress hyperglycemia is common in critically ill patients and is associated with poor prognosis. Whether this association exists in pulmonary hypertension (PH) patients is unknown. The present cohort study investigated the association of stress hyperglycemia with 90-day all-cause mortality in intensive care unit (ICU) patients with PH. Data of the study population were extracted from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. A new index, the ratio of admission glucose to HbA1c (GAR), was used to evaluate stress hyperglycemia. The study population was divided into groups according to GAR quartiles (Q1-Q4). The outcome of interest was all-cause mortality within 90 days, which was considered a short-term prognosis. A total of 53,569 patients were screened. Ultimately, 414 PH patients were enrolled; 44.2% were male, and 23.2% were admitted to the cardiac ICU. As the GAR increased from Q2 to Q4, the groups had lower creatinine levels, longer ICU stays, and a higher proportion of renal disease. After adjusting for confounding factors such as demographics, vital signs, and comorbidities, an elevated GAR was associated with an increased risk of 90-day mortality. Stress hyperglycemia assessed by the GAR was associated with increased 90-day mortality in ICU patients with PH.

Clinical success of anti-infective combination therapy compare to monotherapy in patients with carbapenem-resistant Pseudomonas aeruginosa infection: a 10-years retrospective study

BackgroundCarbapenem-resistant Pseudomonas aeruginosa (CRPA) infection has become a major public health concern. The recommendations for monotherapy and combination therapy in the current guidelines lack sufficient evidence to support them. The primary objective of this study is to determine the effectiveness of anti-Infective combination therapy compared to monotherapy in achieving clinical success in patients with CRPA infection and risk factors of clinical failure of monotherapy.MethodsA retrospective study from Medical Information Mart for Intensive Care IV (MIMIC-IV) was conducted. We included adults with infections caused by CRPA. The outcomes of this study were clinical success, complete clinical success, and 28-day all-cause mortality.ResultsA total of 279 subjects were finally enrolled. The rate of clinical success for combination therapy was higher than that for monotherapy (73.1% versus 60.4%, p=0.028). Compared to clinical failure patients, patients in the clinical success group were more likely to die within 28 days after CRPA was found (48.3% versus 3.6%, p<0.001). In a multivariate logistic regression analysis, monotherapy was found to be significantly correlated with clinical success (OR, 0.559, 95% CI, 0.321-0.976; p = 0.041).ConclusionCombination therapy is more effective for CRPA infection patients, especially those whose SOFA score is ≥ 2 or whose Charlson comorbidity index is ≥ 6.

Association between inflammation markers and all-cause mortality in critical ill patients with atrial fibrillation: Analysis of the Multi-Parameter Intelligent Monitoring in Intensive Care (MIMIC-IV) database

BackgroundInflammation is related to cardiovascular disease. Among the many inflammatory markers, neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and systemic immune-inflammatory index (SII) were considered as novel predictors for atherosclerosis outcomes. We aimed to investigate the impact of these inflammatory markers on the prognosis of patients with atrial fibrillation (AF). MethodsWe obtained data on AF patients from the Medical Information Mart for Intensive Care (MIMIC)-IV database. These patients were classified into two groups based on their survival status within 30 days. Then, they were divided into three groups based on the tertile of baseline NLR, PLR, and SII, respectively. We comprehensively explored the relationship between those inflammatory indicators and all-cause mortality in patients with AF by Kaplan-Meier analysis, multivariate Cox regression analysis, receiver operating characteristic (ROC) analyses, restricted cubic spline regression (RCS), and subgroup analysis. ResultsA total of 4562 patients with AF were included. Statistically significant differences were found between survivor and non-survivor groups for NLR, PLR and SII. Patients in the high tertile of the NLR had a higher mortality rate than those in the low and intermediate tertiles, as did patients in the PLR and the SII. NLR, PLR and SII were independently associated with increased risk of all-cause mortality. RCS showed that the 30-day and 365-day risk of death were linearly associated with increases in NLR, PLR, and SII, respectively. ConclusionNLR, PLR, and SII have the potential to be used as indicators for stratifying the risk of mortality in critically ill patients with AF.

Improved prediction of sepsis-associated encephalopathy in intensive care unit sepsis patients with an innovative nomogram tool.

Sepsis-associated encephalopathy (SAE) occurs as a result of systemic inflammation caused by sepsis. It has been observed that the majority of sepsis patients experience SAE while being treated in the intensive care unit (ICU), and a significant number of survivors continue suffering from cognitive impairment even after recovering from the illness. The objective of this study was to create a predictive nomogram that could be used to identify SAE risk factors in patients with ICU sepsis. We conducted a retrospective cohort study using the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. We defined SAE as a Glasgow Coma Scale (GCS) score of 15 or less, or delirium. The patients were randomly divided into training and validation cohorts. We used least absolute shrinkage and selection operator (LASSO) regression modeling to optimize feature selection. Independent risk factors were determined through a multivariable logistic regression analysis, and a prediction model was built. The performance of the nomogram was evaluated using various metrics including the area under the receiver operating characteristic curve (AUC), calibration plots, Hosmer-Lemeshow test, decision curve analysis (DCA), net reclassification improvement (NRI), and integrated discrimination improvement (IDI). Among the 4,476 sepsis patients screened, 2,781 (62.1%) developed SAE. In-hospital mortality was higher in the SAE group compared to the non-SAE group (9.5% vs. 3.7%, p < 0.001). Several variables were analyzed, including the patient's age, gender, BMI on admission, mean arterial pressure, body temperature, platelet count, sodium level, and use of midazolam. These variables were used to create and validate a nomogram. The nomogram's performance, assessed by AUC, NRI, IDI, and DCA, was found to be superior to the conventional SOFA score combined with delirium. Calibration plots and the Hosmer-Lemeshow test confirmed the accuracy of the nomogram. The enhanced NRI and IDI values demonstrated that our scoring system outperformed traditional diagnostic approaches. Additionally, the DCA curve indicated the practicality of the nomogram in clinical settings. This study successfully identified autonomous risk factors associated with the emergence of SAE in sepsis patients and utilized them to formulate a predictive model. The outcomes of this investigation have the potential to serve as a valuable clinical resource for the timely detection of SAE in patients.

Blood urea nitrogen to serum albumin ratio is associated with all-cause mortality in patients with AKI: a cohort study.

This study aims to investigate the relationship between blood urea nitrogen to serum albumin ratio (BAR) and all-cause mortality in patients with acute kidney injury (AKI) and evaluate the effect of BAR on the prognosis of AKI. Adult patients with AKI admitted to the ICU in the Medical Information Mart for Intensive Care IV (MIMIC-IV) were selected in a retrospective cohort study. BAR (mg/g) was calculated using initial blood urea nitrogen (mg/dl)/serum albumin (g/dl). According to the BAR, these patients were divided into quartiles (Q1-Q4). Kaplan-Meier analysis was used to compare the mortality of the above four groups. Multivariate Cox regression analysis was used to evaluate the association between BAR and 28-day mortality and 365-day mortality. The receiver operating characteristic (ROC) curve was plotted and the area under the curve (AUC) was calculated, and the subgroup analysis was finally stratified by relevant covariates. A total of 12,125 patients with AKI were included in this study. The 28-day and 365-day mortality rates were 23.89 and 39.07%, respectively. Kaplan-Meier analysis showed a significant increase in all-cause mortality in patients with high BAR (Log-rank p < 0.001). Multivariate Cox regression analysis showed that BAR was an independent risk factor for 28-day mortality (4.32 < BAR≤7.14: HR 1.12, 95% CI 0.97-1.30, p = 0.114; 7.14 < BAR≤13.03: HR 1.51, 95% CI 1.31-1.75, p < 0.001; BAR>13.03: HR 2.07, 95% CI 1.74-2.47, p < 0.001; Reference BAR≤4.32) and 365-day mortality (4.32 < BAR≤7.14: HR 1.22, 95% CI 1.09-1.36, p < 0.001; 7.14 < BAR≤13.03: HR 1.63, 95% CI 1.46-1.82, p < 0.001; BAR>13.03: HR 2.22, 95% CI 1.93-2.54, p < 0.001; Reference BAR ≤ 4.32) in patients with AKI. The AUC of BAR for predicting 28-day mortality and 365-day mortality was 0.649 and 0.662, respectively, which is better than that of blood urea nitrogen and sequential organ failure assessment. In addition, subgroup analysis showed a stable relationship between BAR and adverse outcomes in patients with AKI. BAR is significantly associated with increased all-cause mortality in patients with AKI. This finding suggests that BAR may help identify people with AKI at high risk of mortality.

The optimal blood glucose is significantly associated with lower mortality in critically ill patients with cardiogenic shock: an analysis revealed with time series blood glucose records

BackgroundThe optimal blood glucose (BG) level for patients with cardiogenic shock in the intensive care unit (ICU) remains unclear. Studies have found that both excessively high and low BG levels contribute to adverse cardiovascular events. Our study aims to investigate the optimal BG level for critically ill patients with cardiogenic shock and evaluate the effects of optimal BG on the prognosis of patients.MethodsA total of 2013 patients with cardiogenic shock obtained from the Medical Information Mart for Intensive Care (MIMIC) IV database were included in the final cohort for our retrospective observational study for data analysis. The exposure was time-weighted average BG (TWA-BG), which was calculated by the time-series BG records and corresponding time stamps of patients with cardiogenic shock during their stay in the ICU. The cut-off value of TWA-BG was identified by the restricted cubic spline curve and included patients were categorized into three groups: low TWA-BG group (TWA-BG ≤ 104 mg/dl), optimal TWA-BG group (104 < TWA-BG ≤ 138 mg/dl), and high TWA-BG group (TWA-BG > 138 mg/dl). The primary outcome was 28-day mortality, and the secondary outcomes were ICU and in-hospital mortality. We performed the log-rank test to detect whether there is a difference in mortality among different groups in the original cohort. Multiple distinct models were employed to validate the robustness of the results.ResultsOur study revealed that the optimal BG level for critically ill patients with cardiogenic shock is 104–138 mg/dl. Compared to the optimal TWA-BG group, the low TWA-BG group (hazard ratio (HR): 1.67, 95% confidence interval (CI): 1.19–2.33, p = 0.002) and high TWA-BG group (HR: 1.72, 95% CI: 1.46–2.03, p < 0.001) exhibited higher 28-day mortality. Similarly, the low TWA-BG group and high TWA-BG group demonstrated higher risks in terms of ICU mortality (low TWA-BG group: HR: 2.30, 95% CI: 1.40–3.79, p < 0.001; high TWA-BG group: HR: 1.77, 95% CI: 1.45–2.17, p < 0.001) and in-hospital mortality (low TWA-BG group: HR: 1.73, 95% CI: 1.19–2.51, p = 0.001; high TWA-BG group: HR: 1.64, 95% CI: 1.38–1.95, p < 0.001). Sensitivity analysis conducted through propensity score matching and the subgroup analysis further substantiated the robustness of the results.ConclusionThe optimal BG level for patients with cardiogenic shock is 104–138 mg/dl. BG levels below 104 mg/dl and above 138 mg/dl were associated with a less favorable prognosis.

Association between estimated pulse wave velocity and in-hospital mortality of patients with acute kidney injury: a retrospective cohort analysis of the MIMIC-IV database

Background Estimated pulse wave velocity (ePWV) has been found to be an independent predictor of cardiovascular mortality and kidney injury, which can be estimated noninvasively. This study aimed to investigate the association between ePWV and in-hospital mortality in critically ill patients with acute kidney injury (AKI). Methods This study included 5960 patients with AKI from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. The low and high ePWV groups were compared using a Kaplan-Meier survival curve to evaluate the differences in survival status. Cox proportional hazards models were used to explore the association between ePWV and in-hospital mortality in critically ill patients with AKI. To further examine the dose-response relationship, we used a restricted cubic spline (RCS) model. Stratification analyses were conducted to investigate the effect of ePWV on hospital mortality across various subgroups. Results Survival analysis indicated that patients with high ePWV had a lower survival rate than those with low ePWV. Following adjustment, high ePWV demonstrated a statistically significant association with an increased risk of in-hospital mortality among AKI patients (HR = 1.53, 95% CI = 1.36-1.71, p < 0.001). Analysis using the RCS model confirmed a linear increase in the risk of hospital mortality as the ePWV values increased (P for nonlinearity = 0.602). Conclusions A high ePWV was significantly associated with an increased risk of in-hospital mortality among patients with AKI. Furthermore, ePWV was an independent predictor of in-hospital mortality in critically ill patients with AKI.

Improving Prediction of Survival for Extremely Premature Infants Born at 23 to 29 Weeks Gestational Age in the Neonatal Intensive Care Unit: Development and Evaluation of Machine Learning Models.

Infants born at extremely preterm gestational ages are typically admitted to the neonatal intensive care unit (NICU) after initial resuscitation. The subsequent hospital course can be highly variable, and despite counseling aided by available risk calculators, there are significant challenges with shared decision-making regarding life support and transition to end-of-life care. Improving predictive models can help providers and families navigate these unique challenges. Machine learning methods have previously demonstrated added predictive value for determining intensive care unit outcomes, and their use allows consideration of a greater number of factors that potentially influence newborn outcomes, such as maternal characteristics. Machine learning-based models were analyzed for their ability to predict the survival of extremely preterm neonates at initial admission. Maternal and newborn information was extracted from the health records of infants born between 23 and 29 weeks of gestation in the Medical Information Mart for Intensive Care III (MIMIC-III) critical care database. Applicable machine learning models predicting survival during the initial NICU admission were developed and compared. The same type of model was also examined using only features that would be available prepartum for the purpose of survival prediction prior to an anticipated preterm birth. Features most correlated with the predicted outcome were determined when possible for each model. Of included patients, 37 of 459 (8.1%) expired. The resulting random forest model showed higher predictive performance than the frequently used Score for Neonatal Acute Physiology With Perinatal Extension II (SNAPPE-II) NICU model when considering extremely preterm infants of very low birth weight. Several other machine learning models were found to have good performance but did not show a statistically significant difference from previously available models in this study. Feature importance varied by model, and those of greater importance included gestational age; birth weight; initial oxygenation level; elements of the APGAR (appearance, pulse, grimace, activity, and respiration) score; and amount of blood pressure support. Important prepartum features also included maternal age, steroid administration, and the presence of pregnancy complications. Machine learning methods have the potential to provide robust prediction of survival in the context of extremely preterm births and allow for consideration of additional factors such as maternal clinical and socioeconomic information. Evaluation of larger, more diverse data sets may provide additional clarity on comparative performance.

Correlation between early intracranial pressure and cerebral perfusion pressure with 28-day intensive care unit mortality in patients with hemorrhagic stroke.

Hemorrhagic stroke may cause changes in intracranial pressure (ICP) and cerebral perfusion pressure (CPP), which may influence the prognosis of patients. The aim of this study was to investigate the relationship between early ICP, CPP, and 28-day mortality in the intensive care unit (ICU) of patients with hemorrhagic stroke. A retrospective study was performed using the Medical Information Mart for Intensive Care (MIMIC-IV) and the eICU Collaborative Research Database (eICU-CRD), including hemorrhagic stroke patients in the ICU with recorded ICP monitoring. The median values of ICP and CPP were collected for the first 24 h of the patient's monitoring. The primary outcome was 28-day ICU mortality. Multivariable Cox proportional hazards models were used to analyze the relationship between ICP, CPP, and 28-day ICU mortality. Restricted cubic regression splines were used to analyze nonlinear relationships. The study included 837 patients with a 28-day ICU mortality rate of 19.4%. Multivariable analysis revealed a significant correlation between early ICP and 28-day ICU mortality (HR 1.08, 95% CI 1.04-1.12, p < 0.01), whereas early CPP showed no correlation with 28-day ICU mortality (HR 1.00, 95% CI 0.98-1.01, p = 0.57), with a correlation only evident when CPP < 60 mmHg (HR 1.99, 95% CI 1.14-3.48, p = 0.01). The study also identified an early ICP threshold of 16.5 mmHg. Early ICP shows a correlation with 28-day mortality in hemorrhagic stroke patients, with a potential intervention threshold of 16.5 mmHg. In contrast, early CPP showed no correlation with patient prognosis.

Prognostic impacts of repeated sepsis in intensive care unit on autoimmune disease patients: a retrospective cohort study

BackgroundAutoimmune diseases (ADs) may be complicated by sepsis when intensive care unit (ICU) admission. But repeated sepsis among AD patients has not been studied yet. The aim of this study is to investigate the impact of repeated in-ICU sepsis on the 1-year overall-cause mortality, septic shock and in-ICU death of AD patients.MethodsData of AD patients with sepsis retrieved from Medical Information Mart for Intensive Care IV (MIMIC-IV) database were divided into the single group and the repeated group according to the frequency of in-ICU sepsis. Propensity score matching was used to balance inter-group bias. Cox proportional hazard regression and sensitivity analysis were utilized to assess the variables on mortality.ResultsThe incidence of repeated in-ICU sepsis in baseline was 19.8%. The repeated in-ICU sepsis was a risk factor for 1-year overall-cause mortality among AD patients (adjusted hazard ratio [HR] = 1.50, 95% CI: 1.16–1.93, P = 0.002), with robust adjusted HRs by the adjustment for confounders in the sensitivity analysis (all P < 0.01). Maximum Sequential Organ Failure Assessment (Max SOFA), Charlson comorbidity index (CCI) and Simplified Acute Physiology Score-II (SAPS-II) were risk factors for 1-year overall-cause mortality among AD with repeated sepsis (Max SOFA: HR = 1.09, P = 0.002; CCI: HR = 1.08, P = 0.039; SAPS-II: HR = 1.03, P < 0.001).ConclusionsCompared to single hit, repeated in-ICU sepsis was independently related to a higher risk of 1-year overall-cause mortality among AD patients. Assessment tools (Higher SOFA, CCI and SAPS-II scores) were closely linked to poor prognosis of AD with repeated sepsis and helped to reflect ill physical conditions for the patients.

The association between triglyceride glucose-body mass index and all-cause mortality in critically ill patients with atrial fibrillation: a retrospective study from MIMIC-IV database

BackgroundThe TyG-BMI index, which is a reliable indicator of insulin resistance (IR), has been found to have a significant correlation with the occurrence of cardiovascular events. However, there still lacks study on the TyG-BMI index and prognosis in patients with atrial fibrillation (AF). The objective of the present study was to evaluate the relationship between TyG-BMI index at admission to ICU and all-cause mortality in critically ill patients with AF.MethodsThe patient’s data were extracted from Medical Information Mart for Intensive Care IV(MIMIC-IV) database. All patients were divided into four groups according to TyG-BMI index. Outcomes include primary and secondary endpoints, with the primary endpoint being the 30-day and 365-day all-cause mortality and the secondary endpoint being the 90-day and 180-day all-cause mortality. TyG-BMI index was quartile and Kaplan-Meier curve was used to compare the outcome of each group. Cox proportional-hazards regression model and restricted cubic splines (RCS) were conducted to assess the relationship between TyG-BMI index and outcomes.ResultsOut of a total of 2509 participants, the average age was 73.26 ± 11.87 years, with 1555 (62.0%) being males. Patients with lower level of TyG-BMI had higher risk of 30-day, 90-day, 180-day and 365-day all-cause mortality, according to the Kaplan-Meier curves (log-rank P < 0.001). In addition, cox proportional-hazards regression analysis revealed that the risk of 30-day, 90-day, 180-day and 365-day all-cause mortality was significantly higher in the lowest quartile of TyG-BMI. Meanwhile, the RCS analysis indicated that L-typed relationships between TyG-BMI index and all-cause mortality, with inflection points at 223.60 for 30-day and 255.02 for 365-day all-cause mortality, respectively. Compared to patients with TyG-BMI levels below the inflection points, those with higher levels had a 1.8% lower risk for 30-day all-cause mortality (hazard ratio [HR] 0.982, 95% confidence interval [CI] 0.9676–0.988) and 1.1% lower risk for 365-day all-cause mortality (HR 0.989, 95% CI 0.986–0.991).ConclusionIn critically ill patients with AF, a lower TyG-BMI level is significantly associated with a higher risk of 30-day, 90-day, 180-day and 365-day all-cause mortality. TyG-BMI index could be used as a valid indicator for grading and treating patients with AF in the ICU.

Using natural language processing for automated classification of disease and to identify misclassified ICD codes in cardiac disease

Abstract Aims ICD codes are used for classification of hospitalizations. The codes are used for administrative, financial, and research purposes. It is known, however, that errors occur. Natural language processing (NLP) offers promising solutions for optimizing the process. To investigate methods for automatic classification of disease in unstructured medical records using NLP and to compare these to conventional ICD coding. Methods and results Two datasets were used: the open-source Medical Information Mart for Intensive Care (MIMIC)-III dataset (n = 55.177) and a dataset from a hospital in Belgium (n = 12.706). Automated searches using NLP algorithms were performed for the diagnoses ‘atrial fibrillation (AF)’ and ‘heart failure (HF)’. Four methods were used: rule-based search, logistic regression, term frequency-inverse document frequency (TF-IDF), Extreme Gradient Boosting (XGBoost), and Bio-Bidirectional Encoder Representations from Transformers (BioBERT). All algorithms were developed on the MIMIC-III dataset. The best performing algorithm was then deployed on the Belgian dataset. After preprocessing a total of 1438 reports was retained in the Belgian dataset. XGBoost on TF-IDF matrix resulted in an accuracy of 0.94 and 0.92 for AF and HF, respectively. There were 211 mismatches between algorithm and ICD codes. One hundred and three were due to a difference in data availability or differing definitions. In the remaining 108 mismatches, 70% were due to incorrect labelling by the algorithm and 30% were due to erroneous ICD coding (2% of total hospitalizations). Conclusion A newly developed NLP algorithm attained a high accuracy for classifying disease in medical records. XGBoost outperformed the deep learning technique BioBERT. NLP algorithms could be used to identify ICD-coding errors and optimize and support the ICD-coding process.

Machine learning-based models for prediction of the risk of stroke in coronary artery disease patients receiving coronary revascularization.

To construct several prediction models for the risk of stroke in coronary artery disease (CAD) patients receiving coronary revascularization based on machine learning methods. In total, 5757 CAD patients receiving coronary revascularization admitted to ICU in Medical Information Mart for Intensive Care IV (MIMIC-IV) were included in this cohort study. All the data were randomly split into the training set (n = 4029) and testing set (n = 1728) at 7:3. Pearson correlation analysis and least absolute shrinkage and selection operator (LASSO) regression model were applied for feature screening. Variables with Pearson correlation coefficient<9 were included, and the regression coefficients were set to 0. Features more closely related to the outcome were selected from the 10-fold cross-validation, and features with non-0 Coefficent were retained and included in the final model. The predictive values of the models were evaluated by sensitivity, specificity, area under the curve (AUC), accuracy, and 95% confidence interval (CI). The Catboost model presented the best predictive performance with the AUC of 0.831 (95%CI: 0.811-0.851) in the training set, and 0.760 (95%CI: 0.722-0.798) in the testing set. The AUC of the logistic regression model was 0.789 (95%CI: 0.764-0.814) in the training set and 0.731 (95%CI: 0.686-0.776) in the testing set. The results of Delong test revealed that the predictive value of the Catboost model was significantly higher than the logistic regression model (P<0.05). Charlson Comorbidity Index (CCI) was the most important variable associated with the risk of stroke in CAD patients receiving coronary revascularization. The Catboost model was the optimal model for predicting the risk of stroke in CAD patients receiving coronary revascularization, which might provide a tool to quickly identify CAD patients who were at high risk of postoperative stroke.

A comparison of invasive arterial blood pressure measurement with oscillometric non-invasive blood pressure measurement in patients with sepsis.

This study aimed to compare non-invasive oscillometric blood pressure (NIBP) measurement with invasive arterial blood pressure (IBP) measurement in patients with sepsis. We conducted a retrospective study to evaluate the agreement between IBP and NIBP using the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. Paired blood pressure measurements of mean arterial pressure (MAP), systolic blood pressure (SBP), and diastolic blood pressure (DBP) were compared using Bland-Altman analysis and paired Student's t test. We also focus on the effect of norepinephrine (NE) on the agreement between the two methods and the association between blood pressure and mortality during intensive care unit (ICU) stay. A total of 96,673 paired blood pressure measurements from 6060 unique patients were analyzed in the study. In Bland-Altman analysis, the bias (± SD, 95% limits of agreement) was 6.21mmHg (± 12.05mmHg, -17.41 to 29.83mmHg) for MAP, 0.39mmHg (± 19.25mmHg, -37.34 to 38.12mmHg) for SBP, and 0.80mmHg (± 12.92mmHg, -24.52 to 26.12mmHg) for DBP between the two techniques. Similarly, large limits of agreement were shown in different groups of NE doses. NE doses significantly affected the agreement between IBP and NIBP. SBP between the two methods gave an inconsistent assessment of patients' risk of ICU mortality. IBP and NIBP were not interchangeable in septic patients. Clinicians should be aware that non-invasive MAP was clinically and significantly underestimated invasive MAP.

Association Between Serum Anion Gap and Mortality in Critically Ill Patients with COPD in ICU: Data from the MIMIC IV Database.

Serum anion gap (AG) has been proven to be associated with prognosis in critically ill patients. However, few studies have investigated the association between AG and all-cause mortality in critically ill patients with chronic obstructive pulmonary disease (COPD). We hypothesized that the initial AG level would predict the mortality risk in critically ill patients with COPD. This retrospective cohort study was based on the Medical Information Mart for Intensive Care (MIMIC) IV database. We extracted demographics, vital signs, laboratory tests, comorbidity, and scoring systems from the first 24 hours after patient ICU admission. Multivariable logistic regression analysis models were used to explore the association between serum AG levels and mortality. Interaction and stratified analyses were conducted including age, gender and comorbidity. A total of 5531 critically ill patients with COPD were enrolled, composed of 53.6% male and 46.4% female with a median age of 73 years. The all-cause mortality of these patients during ICU hospitalization was 13.7%. The risk of all-cause mortality increased as the AG level increased in the univariate logistic regression analysis (OR=1.13, 95% CI: 1.11-1.15, p<0.01). After adjusting for all the covariates in multivariate logistic regression analysis, the odds ratio was 1.06 (95% CI: 1.04-1.09, p<0.01). Compared with the lowest AG group Q1 (≤11mmol/L), the adjusted OR value for AG and mortality in Q2 (12-13mmol/L) was 0.89 (95% CI: 0.63-1.25, p=0.502), Q3 (14-15mmol/L) was 0.95 (95% CI: 0.68-1.34, p=0.788), and Q4 (≥16mmol/L) was 1.49 (95% CI: 1.10-2.02, p=0.009) respectively. In addition, the results of the subgroup and stratified analyses were robust. AG is positively related to all-cause mortality in critically ill patients with COPD.

The prediction of in-hospital mortality in elderly patients with sepsis-associated acute kidney injury utilizing machine learning models

BackgroundSepsis-associated acute kidney injury (SA-AKI) is a severe complication associated with poorer prognosis and increased mortality, particularly in elderly patients. Currently, there is a lack of accurate mortality risk prediction models for these patients in clinic. ObjectivesThis study aimed to develop and validate machine learning models for predicting in-hospital mortality risk in elderly patients with SA-AKI. MethodsMachine learning models were developed and validated using the public, high-quality Medical Information Mart for Intensive Care (MIMIC)-IV critically ill database. The recursive feature elimination (RFE) algorithm was employed for key feature selection. Eleven predictive models were compared, with the best one selected for further validation. Shapley Additive Explanations (SHAP) values were used for visualization and interpretation, making the machine learning models clinically interpretable. ResultsThere were 16,154 patients with SA-AKI in the MIMIC-IV database, and 8426 SA-AKI patients were included in this study (median age: 77.0 years; female: 45%). 7728 patients excluded based on these criteria. They were randomly divided into a training cohort (5,934, 70%) and a validation cohort (2,492, 30%). Nine key features were selected by the RFE algorithm. The CatBoost model achieved the best performance, with an AUC of 0.844 in the training cohort and 0.804 in the validation cohort. SHAP values revealed that AKI stage, PaO2, and lactate were the top three most important features contributing to the CatBoost model. ConclusionWe developed a model capable of predicting the risk of in-hospital mortality in elderly patients with SA-AKI.