Predictors Of Hospital Mortality Research Articles

Prediction of hospital mortality in intensive care unit patients from clinical and laboratory data: A machine learning approach.

BACKGROUNDIntensive care unit (ICU) patients demand continuous monitoring of several clinical and laboratory parameters that directly influence their medical progress and the staff’s decision-making. Those data are vital in the assistance of these patients, being already used by several scoring systems. In this context, machine learning approaches have been used for medical predictions based on clinical data, which includes patient outcomes.AIMTo develop a binary classifier for the outcome of death in ICU patients based on clinical and laboratory parameters, a set formed by 1087 instances and 50 variables from ICU patients admitted to the emergency department was obtained in the “WiDS (Women in Data Science) Datathon 2020: ICU Mortality Prediction” dataset. METHODSFor categorical variables, frequencies and risk ratios were calculated. Numerical variables were computed as means and standard deviations and Mann-Whitney U tests were performed. We then divided the data into a training (80%) and test (20%) set. The training set was used to train a predictive model based on the Random Forest algorithm and the test set was used to evaluate the predictive effectiveness of the model.RESULTSA statistically significant association was identified between need for intubation, as well predominant systemic cardiovascular involvement, and hospital death. A number of the numerical variables analyzed (for instance Glasgow Coma Score punctuations, mean arterial pressure, temperature, pH, and lactate, creatinine, albumin and bilirubin values) were also significantly associated with death outcome. The proposed binary Random Forest classifier obtained on the test set (n = 218) had an accuracy of 80.28%, sensitivity of 81.82%, specificity of 79.43%, positive predictive value of 73.26%, negative predictive value of 84.85%, F1 score of 0.74, and area under the curve score of 0.85. The predictive variables of the greatest importance were the maximum and minimum lactate values, adding up to a predictive importance of 15.54%.CONCLUSIONWe demonstrated the efficacy of a Random Forest machine learning algorithm for handling clinical and laboratory data from patients under intensive monitoring. Therefore, we endorse the emerging notion that machine learning has great potential to provide us support to critically question existing methodologies, allowing improvements that reduce mortality.

Long-term results after simultaneous carotid and coronary revascularisation.

The revascularisation strategy for concomitant carotid and coronary disease is unknown. Simultaneous or stage coronary artery stenting and carotid endarterectomy are the most common revascularisation approach in the CABG population. This study aimed to evaluate long-term results after simultaneous carotid artery stenting or carotid endarterectomy in patients who underwent coronary artery bypass surgery. This is a prospective cohort non-randomised single-institution study. During the period from 2012 to 2015, sixty consecutive patients (65.9 ± 7.41 mean) underwent simultaneous carotid artery stenting and coronary artery bypass surgery (n = 30) or simultaneous carotid endarterectomy and coronary artery bypass surgery (n = 30). The primary endpoints were short- and long-term rates of adverse events (transient ischemic attack, stroke, myocardial infarction, and death). The mean follow-up was 62.05 ± 11.12 months. In-hospital mortality was insignificantly higher in the carotid endarterectomy, and coronary artery bypass surgery group (6.6% vs. 0%), the rate of stroke and myocardial infarction was similar (13.3% and 0% in the carotid endarterectomy and coronary artery bypass surgery group vs. 6.6% and 3.3% in the carotid artery stenting and coronary artery bypass surgery group, respectively). The intensive care unit readmission was significantly higher in the surgical revascularisation approach; it was an independent predictor of hospital mortality. The overall mortality during the follow-up period was 14.28% in both groups. Freedom of the composite adverse outcomes (stroke, myocardial infarction, and death) was 78.55%. Comparing two revascularisation strategies is not straightforward due to different anatomical indications for carotid artery stenting and endarterectomy. We consider that each technique has an essential role in carotid revascularisation. Good selection of patients, according to indications, contributes to satisfactory short- and long-term results.

Machine learning-based prediction of in-hospital mortality for post cardiovascular surgery patients admitting to intensive care unit: a retrospective observational cohort study based on a large multi-center critical care database

Background and objectivesThe acute physiology and chronic health evaluation-IV model (APACHE-IV), and the sequential organ failure assessment (SOFA) score are two traditional severity assessment systems that can be applied to cardiac surgery patients admitted to intensive care units (ICUs). However, the performance of machine learning approaches in post cardiovascular surgery (PCS) patients admitted to the ICU remains unknown. MethodsThe clinical data of adult subjects were collected from the eICU database. Seven models were constructed based on the training set (70% random sample) for predicting hospital mortality, including two traditional models based on APACHE-IV and SOFA scores and five machine learning models. We measured the models’ performance in the remaining 30% of the sample by computing AUC-ROC values, prospective prediction results, and decision curves and compared the models with net reclassification improvement. ResultsThis study included 5860 PCS patients. The AUC-ROC value of the Xgboost model significantly outperformed the APACHE-IV and SOFA scores (0.12 [0.06–0.17] p < 0.01, 0.18 [0.1–0.26] p < 0.01 respectively). The use of ML models would also gain more clinical net benefits than traditional models based on decision curve analysis. There was a significant improvement in integrated discrimination when comparing the backward stepwise linear regression model with the APACHE-IV model (0.11 [0.05, 0.16], p < 0.01) and SOFA model (0.12 [0.06, 0.17], p < 0.01). ConclusionsIn conclusion, the predictive ability of ML models was better than that of traditional models. The present study suggested that developing advanced prognosis prediction tools could support clinical decision-making in the ICU for PCS patients.

The Role of the Monocyte-to-Lymphocyte Ratio in Acute Ischemic Stroke Patients with Acute Kidney Injury.

Objective The objective of the study was to determine if acute kidney injury (AKI) in patients with acute ischemic stroke was associated with the monocyte-to-lymphocyte ratio (MLR) assessed upon admission to the neurology intensive care unit (NICU) (AIS). We also looked into the MLR's function in predicting hospital mortality in AIS patients. Methods This retrospective analysis included 595 adult patients with AIS who were hospitalized to the NICU of the First Affiliated Hospital of South China between January 2017 and December 2019. Clinical signs and imaging studies were used to diagnose AIS. KDIGO criteria were used to define AKI. The ratio of monocytes to lymphocytes was used to compute MLR, the ratio of neutrophils to lymphocytes was used to calculate NLR, and the ratio of platelets to lymphocytes was used to calculate PLR. Result 361 males and 234 women between the ages of 66.27 ± 12.05 years took part in the study. The individuals' MLR was 0.4729 ± 0.3461 and their neutrophil-to-lymphocyte ratio (NLR) was 8.18 ± 5.45. There were notable disparities in MLR and NLR between the AKI and non-AKI groups (p < 0.001). The link between MLR and AKI development risk was enhanced after adjustment, with respective cutoff values of 0.4581 and 9.26. For the MLR-based prediction of AKI incidence, the areas under the receiver-operating characteristic curves (AUCs) were 0.711 (95% CI: 0.663-0.758). And NLR-based prediction of AKI incidence the AUCs was (95% CI: 0.742-0826). Additionally, MLR was associated with a higher rate of in-hospital mortality (2.825, 95% confidence interval: 1.058, 7.545), whereas NLR was associated with a risk of in-hospital mortality of 1.085. (95 percent CI: 1.022, 1.151). An AUC of 0.745 (95% CI: 0.601-0.889, p = 0.026) was obtained for in-hospital mortality based on the MLR, whereas an AUC of 0.724 (95% CI: 0.531-0.916, p = 0.042) was obtained for in-hospital mortality based on the NLR. Conclusion MLR and neutrophil-to-lymphocyte ratio (NLR) were associated with a higher risk of AKI and in-hospital death in AIS patients.

Association Between Red Blood Cell Distribution Width-Albumin Ratio and Hospital Mortality in Chronic Obstructive Pulmonary Disease Patients Admitted to the Intensive Care Unit: A Retrospective Study.

PurposeHigh levels of red blood cell distribution width (RDW) and hypoalbuminemia are markers of poor prognosis in chronic obstructive pulmonary disease (COPD) patients. However, few studies have shown that the red blood cell distribution width–albumin ratio (RAR) is related to the mortality of COPD. This study aimed to explore the relationship between RAR and hospital mortality in COPD patients admitted to the intensive care unit (ICU).Patients and MethodsPatients were retrospectively incorporated from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database and divided into two groups by a cutoff value of RAR. Propensity score matching (PSM) was performed to adjust for the imbalance of covariates. Logistic regression models and subgroup analyses were carried out to investigate the relationship between RAR and hospital mortality. The receiver operating characteristic (ROC) curve was used to evaluate the predictive performance of RAR and decision curve analysis (DCA) to assess the clinical utility.ResultsIn total, 1174 patients were finally identified from the MIMIC-IV database. The cutoff value for RAR was 5.315%/g/dL. After PSM at a 1:1 ratio, 638 patients were included in the matched cohort. In the original and matched cohorts, the high RAR group had higher hospital mortality and longer hospital stays. Logistic regression analysis suggested that RAR was an independent risk factor for hospital mortality. The areas under the ROC curve in the original and matched cohorts were 0.706 and 0.611, respectively, which were larger than applying RDW alone (the original cohort: 0.600, the matched cohort: 0.514). The DCA indicated that RAR had a clinical utility.ConclusionA higher RAR (>5.315%/g/dL) was associated with hospital mortality in COPD patients admitted to ICU. As an easily available peripheral blood marker, RAR can predict hospital mortality in critically ill patients with COPD independently.

Host response dysregulations amongst adults hospitalized by influenza A H1N1 virus pneumonia: A prospective multicenter cohort study

BackgroundLimited knowledge exists on how early host response impacts outcomes in influenza pneumonia. MethodsThis study assessed what was the contribution of host immune response at the emergency department on hospital mortality amongst adults with influenza A H1N1pdm09 pneumonia and whether early stratification by immune host response anticipates the risk of death. This is a secondary analysis from a prospective, observational, multicenter cohort comparing 75 adults requiring intensive care with 38 hospitalized in medical wards. Different immune response biomarkers within 24 h of hospitalization and their association with hospital mortality were assessed. ResultsFifty-three were discharged alive. Non-survivors were associated (p<0.05) with lower lymphocytes (751 vs. 387), monocytes (450 vs. 220) expression of HLA-DR (1,662 vs. 962) and higher IgM levels (178 vs. 152;p<0.01). Lymphocyte subpopulations amongst non-survivors showed a significantly (p<0.05) lower number of TCD3+ (247.2 vs. 520.8), TCD4+ (150.3 vs. 323.6), TCD8+ (95.3 vs. 151.4) and NKCD56+ (21.9 vs. 91.4). Number of lymphocytes, monocytes and NKCD56+ predicted hospital mortality (AUC 0.854). Hospital mortality was independently associated with low HLA-DR values, low number of NKCD56+ cells, and high IgM levels, in a Cox-proportional hazard analysis. A second model, documented that hospital mortality was independently associated with a phenotype combining immunoparalysis with hyperinflammation (HR 5.53; 95%CI 2.16–14.14), after adjusting by predicted mortality. ConclusionsWe conclude that amongst influenza pneumonia, presence of immunoparalysis was a major mortality driver. Influenza heterogeneity was partly explained by early specific host response dysregulations which should be considered to design personalized approaches of adjunctive therapy.

Dynamics of echocardiographic parameters in patients with severe COVID-19 during hospitalization

BACKGROUND: The severe course of a new coronavirus infection may be associated with cardiovascular damage. Thus, there is a need to search for available methods of cardiac status examination to determine management tactics and immediate prognosis in patients with coronavirus disease 2019 (COVID-19).&#x0D; AIM: This study aimed to assess the dynamics of echocardiographic parameters in patients with severe infection during hospitalization.&#x0D; MATERIALS AND METHODS: Forty patients hospitalized for COVID-19 were included in the study. All patients underwent echocardiography on admission to the hospital and in dynamics, that is, upon deterioration of the underlying disease. The ultrasound picture of the heart was compared with the severity of parenchymal damage based on computed tomography (CT) and the degree of respiratory failure based on blood oxygen saturation by pulse oximetry. Concomitant nosologies and autopsy findings were also analyzed.&#x0D; RESULTS: In dynamics, a significant decrease in left ventricular ejection fraction (LVEF) from 47.0% (40.056.0) to 43.5% (37.049.0; p=0.002) was found in the deceased group, which was not observed in the discharged group. Comparative analysis of this index in the discharged and deceased group in study 2 revealed significantly lower values of LVEF in patients with unfavorable hospital prognosis: 51.0% (44.055.0) and 43.5% (37.049.0), respectively (p=0.049). Furthermore, a significantly higher level of systolic pulmonary artery pressure was observed in the deceased group in study 1: 54.0 (50.059.0) and 60.0 (55.570.0) mmHg, respectively (p=0.02), with increasing differences in dynamics, that is, 65.0 (60.070.0) and 50.0 (46.057.0) mmHg, respectively (p 0.0001). Among the indicators reflecting the state of the right ventricle, right ventricular end-systolic size in study 2 in the deceased group (p=0.062) and more pronounced dilatation by this indicator were noted in the group with unfavorable prognosis compared with the discharged group at re-diagnosis (p=0.071).&#x0D; CONCLUSIONS: Dynamic reduction of baseline low LVEF and progression of pulmonary hypertension are the main echocardiographic predictors of hospital mortality in patients with COVID-19. Increased volume of pulmonary parenchyma lesions on CT determines unfavorable hospital prognosis.

Variation in Bed-to-Physician Ratios During Weekday Daytime Hours in ICUs in Australia and New Zealand.

To determine common "bed-to-physician" ratios during weekday hours across ICUs and assess factors associated with variability in this ratio. Retrospective cohort study. All ICUs in Australia/New Zealand that participated in a staffing survey administered in 2017-2018. ICU admissions from 2016 to 2018. We linked survey data with patient-level data. We defined: 1) bed-to-intensivist ratio as the number of usually available ICU beds divided by the number of onsite weekday daytime intensivists; and 2) bed-to-physician ratio as the number of available ICU beds divided by the total number of physicians (intensivists + nonintensivists, including trainees). We calculated the median and interquartile range (IQR) of bed-to-intensivist ratio and bed-to-physician ratios during weekday hours. We assessed variability in each by type of hospital and ICU and by severity of illness of patients, defined by the predicted hospital mortality. None. Of the 123 (87.2%) of Australia/New Zealand ICUs that returned staffing surveys, 114 (92.7%) had an intensivist present during weekday daytime hours, and 116 (94.3%) reported at least one nonintensivist physician. The median bed-to-intensivist ratio was 8.0 (IQR, 6.0-11.4), which decreased to a bed-to-physician ratio of 3.0 (IQR, 2.2-4.9). These ratios varied with mean severity of illness of the patients in the unit. The median bed-to-intensivist ratio was highest (13.5) for ICUs with a mean predicted mortality > 2-4%, and the median bed-to-physician ratio was highest (5.7) for ICUs with a mean predicted mortality of > 4-6%. Both ratios decreased and plateaued in ICUs with a mean predicted mortality for patients greater than 8% (median bed-to-intensivist ratio range, 6.8-8.0, and bed-to-physician ratio range of 2.4-2.7). Weekday bed-to-physician ratios in Australia/New Zealand ICUs are lower than the bed-to-intensivist ratios and have a relatively fixed ratio of less than 3 for units taking care of patients with a higher average severity of illness. These relationships may be different in other countries or healthcare systems.

Doppler-derived haemodynamics performed during admission echocardiography predict in-hospital mortality in cardiac intensive care unit patients.

Cardiac point-of-care ultrasound (CV-POCUS) has become a fundamental part for the assessment of patients admitted to cardiac intensive care units (CICU). We sought to refine the practice of CV-POCUS by identifying 2D and Doppler-derived measurements from bedside transthoracic echocardiograms (TTEs) performed in the CICU that are associated with mortality. We retrospectively included Mayo Clinic CICU patients admitted from 2007 to 2018 and assessed the TTEs performed within 1 day of CICU admission, including Doppler and 2D measurements of left and right ventricular function. Logistic regression and classification and regression tree (CART) analysis were used to determine the association between TTE variables with in-hospital mortality. A total of 6957 patients were included with a mean age of 68.0 ± 14.9 years (37.0% females). A total of 609 (8.8%) patients died in the hospital. Inpatient deaths group had worse biventricular systolic function [left ventricular ejection fraction (LVEF) 48.2 ± 16.0% vs. 38.7 ± 18.2%, P < 0.0001], higher filling pressures, and lower forward flow. The strongest TTE predictors of hospital mortality were left ventricular outflow tract velocity-time integral [LVOT VTI, adjusted OR 0.912 per 1 cm higher, 95% confidence interval (CI) 0.883-0.942, P < 0.0001] followed by medial mitral E/e' ratio (adjusted OR 1.024 per 1 unit higher, 95% CI 1.010-1.039, P = 0.0011). Classification and regression tree analysis identified LVOT VTI <16 cm as the most important TTE predictor of mortality. Doppler-derived haemodynamic TTE parameters have a strong association with mortality in the CICU, particularly LVOT VTI <16 cm or mitral E/e' ratio >15. The incorporation of these simplified Doppler-derived haemodynamics into admission CV-POCUS facilitates early risk stratification and strengthens the clinical yield of the ultrasound exam.

Prediction of Mortality and Major Adverse Kidney Events in Critically Ill Patients With Acute Kidney Injury

Risk prediction tools for assisting acute kidney injury (AKI) management have focused on AKI onset but have infrequently addressed kidney recovery. We developed clinical models for risk stratification of mortality and major adverse kidney events (MAKE) in critically ill patients with incident AKI. Multicenter cohort study. 9,587 adult patients admitted to heterogeneous intensive care units (ICUs; March 2009 to February 2017) who experienced AKI within the first 3 days of their ICU stays. Multimodal clinical data consisting of 71 features collected in the first 3 days of ICU stay. (1) Hospital mortality and (2) MAKE, defined as the composite of death during hospitalization or within 120 days of discharge, receipt of kidney replacement therapy in the last 48 hours of hospital stay, initiation of maintenance kidney replacement therapy within 120 days, or a≥50% decrease in estimated glomerular filtration rate from baseline to 120 days from hospital discharge. Four machine-learning algorithms (logistic regression, random forest, support vector machine, and extreme gradient boosting) and the SHAP (Shapley Additive Explanations) framework were used for feature selection and interpretation. Model performance was evaluated by 10-fold cross-validation and external validation. One developed model including 15 features outperformed the SOFA (Sequential Organ Failure Assessment) score for the prediction of hospital mortality, with areas under the curve of 0.79 (95% CI, 0.79-0.80) and 0.71 (95% CI, 0.71-0.71) in the development cohort and 0.74 (95% CI, 0.73-0.74) and 0.71 (95% CI, 0.71-0.71) in the validation cohort (P< 0.001 for both). A second developed model including 14 features outperformed KDIGO (Kidney Disease: Improving Global Outcomes) AKI severity staging for the prediction of MAKE: 0.78 (95% CI, 0.78-0.78) versus 0.66 (95% CI, 0.66-0.66) in the development cohort and 0.73 (95% CI, 0.72-0.74) versus 0.67 (95% CI, 0.67-0.67) in the validation cohort (P< 0.001 for both). The models are applicable only to critically ill adult patients with incident AKI within the first 3 days of an ICU stay. The reported clinical models exhibited better performance for mortality and kidney recovery prediction than standard scoring tools commonly used in critically ill patients with AKI in the ICU. Additional validation is needed to support the utility and implementation of these models. Acute kidney injury (AKI) occurs commonly in critically ill patients admitted to the intensive care unit (ICU) and is associated with high morbidity and mortality rates. Prediction of mortality and recovery after an episode of AKI may assist bedside decision making. In this report, we describe the development and validation of a clinical model using data from the first 3 days of an ICU stay to predict hospital mortality and major adverse kidney events occurring as long as 120 days after hospital discharge among critically ill adult patients who experienced AKI within the first 3 days of an ICU stay. The proposed clinical models exhibited good performance for outcome prediction and, if further validated, could enable risk stratification for timely interventions that promote kidney recovery.

Predictors of in-hospital mortality in a cohort of people living with HIV (PLHIV) admitted to an academic medical intensive care unit from 2009 to 2014: A retrospective cohort study.

Outcomes for critically ill people living with human immunodeficiency virus (PLHIV) have changed with the use of antiretroviral therapy (ART). To identify these outcomes and correlates of mortality in a contemporary critically ill cohort in an urban academic medical center in Baltimore, a city with a high burden of HIV, we conducted a retrospective cohort study of individuals admitted to a medical intensive care unit (MICU) at a tertiary care center between 2009 and 2014. PLHIV who were at least 18 years of age with an index MICU admission of ≥24 hours during the 5-year study period were included in this analysis. Data were obtained for participants from the time of MICU admission until hospital discharge and up to 180 days after MICU admission. Logistic regression was used to identify independent predictors of hospital mortality. Between June 2009 and June 2014, 318 PLHIV admitted to the MICU met inclusion criteria. Eighty-six percent of the patients were non-Hispanic Blacks. Poorly controlled HIV was very common with 70.2% of patients having a CD4 cell count <200 cells/mm3 within 3 months prior to admission and only 34% of patients having an undetectable HIV viral load. Hospital mortality for the cohort was 17%. In a univariate model, mortality did not differ by demographic variables, CD4 cell count, HIV viral load, or ART use. Regression analysis adjusted by relevant covariates revealed that MICU patients admitted from the hospital ward were 6.4 times more likely to die in hospital than those admitted from emergency department. Other positive predictors were a diagnosis of end-stage liver disease, cardiac arrest, ventilator-dependent respiratory failure, vasopressor requirement, non-Hodgkin lymphoma, and symptomatic cytomegalovirus disease. In conclusion, in this critically ill cohort with HIV infection, most predictors of mortality were not directly related to HIV and were similar to those for the general population.

Acil Serviste Pulmoner Emboli Tanısı Alan Hastalarda Laktat ve Laktat Klerensinin Mortalite Üzerine Etkisinin Retrospektif Değerlendirilmesi

Aim: Pulmonary embolism (PE) is the third leading cause of death due to cardiovascular disease. The presence of shock or hypotension remains the main prognostic clinical marker and to date, it is the only factor that indicates the need for aggressive treatment. Most patients with pulmonary embolism are normotensive. Prognostic indicators are needed to better classify patients with pulmonary embolism. This study aimed to investigate the effect of lactate level on predicting hospital mortality in patients diagnosed with pulmonary embolism in the emergency department, and the effect of lactate clearance on mortality in patients with high lactate levels (lactate ≥2 mmol/L). Material and Methods: This study was designed as a retrospective study. Adult patients (&amp;gt;18-years old) who were diagnosed with PE by computed tomography angiography of the thorax in the university hospital emergency between January 1, 2018, and December 31, 2019, were analyzed. Results: A total of 367 patients were diagnosed with pulmonary embolism in the emergency department. In-hospital mortality rate of the patients was 29.8% and the pulmonary embolism-related mortality rate was 12.4%. Lactate clearance was not significant in predicting mortality risk. In the logistic regression analysis performed among the risk factors affecting mortality in patients with pulmonary embolism, lactate ≥3 mmol/L and normotensive + lactate ≥3 mmol/L were significant in predicting in-hospital mortality. Conclusion: The lactate level in the emergency department could be an effective screening method for identifying mortality in acute pulmonary embolism patients.

Neutrophil-to-Lymphocyte Ratio Predicts in-Hospital Mortality in Intracerebral Hemorrhage

ObjectivesThe neutrophil-to-lymphocyte ratio is gaining popularity as a low-cost biomarker of inflammation and outcome prediction. Intracerebral hemorrhage has high mortality and disability, which strongly influences societal development. This study aimed to research whether the neutrophil-to-lymphocyte ratio can predict hospital mortality in patients with severe intracerebral hemorrhage. Materials and methodsIt was a retrospective analysis of prospectively collected data. The survival analysis and proportional hazards models analyzed clinical data from the Medical Information Mart for Intensive Care III database of patients with intracerebral hemorrhage. ResultsThe records of 1,000 patients were included in our study. Two hundred forty-seven individuals died in the hospital, while 753 survived to discharge. According to data analysis, the neutrophil-to-lymphocyte ratio of the death group (11.21±7.81) significantly exceeded the survival group (7.94±6.04). The univariate Cox regression revealed that the neutrophil-to-lymphocyte ratio is a potential predictor of in-hospital mortality (HR:1.044; 95% CI:1.029-1.059; p <0.001). Furthermore, the proportional hazards model demonstrated that the risk of in-hospital death increased 2.34-fold for each increase in neutrophil-to-lymphocyte ratio when other factors were held constant. Following the ROC analysis, the Kaplan-Meier based on the proportional hazards model showed that patients with a neutrophil-to-lymphocyte ratio >7.68 on the first day of hospitalization had a higher risk of death. ConclusionsIn patients with severe intracerebral hemorrhage, the neutrophil-to-lymphocyte ratio is a potential predictor of in-hospital mortality.

PB2272: THERAPY AND HOSPITAL MORTALITY PREDICTORS IN HEMATOLOGICAL PATIENTS WITH 4 GRADE NEUTROPENIA AND CONCOMITANT CORONAVIRUS INFECTION

Background: According to published data, cancer patients are 5 times more likely to contract coronavirus infection (COVID-19) than non-cancer patients. Aims: To assess and identify predictors of hospital mortality in hematological patients with grade 4 neutropenia and concomitant coronavirus infection. Methods: A prospective study included 130 patients (55% women, 45% men) aged 18 to 91 years (median 59) with acute leukemia (AL) and lymphoproliferative diseases (LPD) who had grade 4 neutropenia and coronavirus infection (COVID-19) and were observed from April 2020 to December 2021 at Moscow City Clinical Hospital 52 (Russia). The study also included a reference group of 130 patients matched by sex, age and diagnosis, with AL/LPD and COVID-19, but without grade 4 neutropenia. The statistical analysis was an intention to treat analysis with adjustment for false discovery rate. Results: AML prevailed in the AL group (80.6%), aggressive non-Hodgkin’s lymphomas in the LPD group. Distribution by disease status: 62 patients (47.7%) on chemotherapy (CT), 19 (14.6%) in remission, 23 (17.8%) with disease progression, 19 (14.6%) with newly diagnosed disease, 7 (5.3%) with a relapse. The patients had a high NEWS score (7.5) and a high Charlson comorbidity index (5). The majority had moderate or severe lung tissue damage (CT2 in 44 (33%), CT3 in 25 (20%), CT4 in 13 (10%)). There were increases in the levels of C-reactive protein (> 100 g/L in 55% of patients, mean 241.5 g/L), LDH (506 U/L), D-dimer (1866), procalcitonin (> 0.5 ng/mL in 73%). IL-6 and IL-1b blockers were used for hypercytokinemia; the dose of IL blockers was reduced in view of grade 4 neutropenia. Empiric antibiotic therapy for complications was used in 91.5% of patients. Antifungal therapy was carried out in 65% of cases. Sepsis developed in 63 (48.5%) patients. The median duration of hospitalization in patients with grade 4 neutropenia was 18 days. Transfer to the ICU was required in 78 (60%) patients in the total group and in 41 patients (52.5%) among those older than 60 years. Mortality in ICU patients: 80.7%, mortality in patients older than 60 years: 92.7%. In non-ICU patients with AL and LPD younger than 60 years, the mortality rate was 0%. Mechanical ventilation was used in 48.5% of cases. Mortality was lower in vaccinated patients than in unvaccinated ones (40% vs 51%). The mortality rates in the main group (AL 48.6%, LPD 51.7%, Fig. 1) and the reference group (AL 37.5%, LPD 13.8%, Fig. 2) showed that the presence of grade 4 neutropenia increased the risk of death almost twofold: the relative risk was 1.9 (95% CI 1.3–2.6), p=0.0001. According to the results of multivariate analysis (logistic regression), the predictors of hospital mortality in the main group were ICU stay (odds ratio 12.1 (1.4–24.0), p=0.0001), sepsis (3.0 (1.1–7.7), p=0.001), and age > 65 (3.0 (0.9–7.5), p=0.001). Severe disease progression/relapse is also a predictor of mortality (close to statistical significance). Image:Summary/Conclusion: Coronavirus infection with severe neutropenia (caused by tumor progression and/or combination chemotherapy) is a significant adverse factor of hospital mortality in patients with hematological cancers. Predictors of hospital mortality are age over 60 years, sepsis, ICU stay, and the absence of remission.

PB1870: COVID - 19 IN PATIENTS WITH CHRONIC LYMPHOCYTIC LEUKEMIA: HOSPITAL MORALITY PREDICTORS

Background: Patients with chronic lymphocytic leukemia (CLL), as the disease progresses and the tumor load increases, are characterized by high susceptibility and severe course of viral infections. The depth of immune dysfunction can be influenced by factors such as age, comorbidity, disease status and previous treatment. The detection of SARS-COV2 in this cohort of patients made it possible to analyze the course of COVID-19, as well as to develop a strategy for the disease management of these patients. Aims: To assess the risk factors for the severe course of COVID-19, as well as to identify predictors of hospital mortality. Methods: 179 patients with verified CLL and COVID-19 were followed up by the hematology service of Moscow City Clinical Hospital No.52 (Moscow) from April 21, 2020 to November 31, 2021. Anthropometric, anamnestic, clinical and laboratory data were evaluated, treatment tactics of both CLL and COVID-19 were analyzed in 179 patients (69 women and 110 men, median age - 66 years). A statistical multivariate analysis of hospital mortality, factors of unfavorable prognosis of COVID-19 (complications, humoral response, hospital mortality) was carried out. Results: Age was a significant adverse factor for men: the age of a man from 73 years and more, increases the risk of death of the patient by 3.0 times (p<0.05). For women, age was not a significant adverse factor. Assessment of the effect of comorbidity: patients with 6 or more comorbidity scores according to Charlson have 2.2 times greater risk of death; the presence of cardiovascular diseases increases the risk of death of the patient by 2.3 times. The presence of 1 or more lines of chemotherapy in the anamnesis increased the risk of hospital mortality by 1.5 times. Patients who had previously undergone chemotherapy were treated until the moment of discharge statistically significantly longer than patients who had not undergone antitumor therapy. Relapse/progression of the disease was a factor of unfavorable prognosis for the survival of patients with CLL (mortality was 47% compared to 14.7% in patients in remission). A significant unfavorable prognosis is also the Binnet stage C (p<0.05). A valid factor in the failure of COVID-19 (p<0.05) was ICU therapy and the development of secondary bacterial complications (the use of 2nd-line antibacterial therapy). The assessment of the humoral response in 59 patients shows the presence of an increase in IgG on the 10th-14th days of the disease in 44% of patients, among whom no deaths were detected, 1/3 were vaccinated patients. The absence of a humoral response was a statistically significant adverse outcome factor in 25% of patients (p<0.05). Summary/Conclusion: Progression/recurrence of chronic lymphocytic leukemia is a factor in the severe course of COVID-19, and, as a result, increases the risk of hospital mortality. The absence of a humoral response determines the persistent course of COVID-19 and increases the risk of death by 25%. The study of the course of COVID-19 in patients with CLL allows us to identify key aspects affecting the choice of management tactics for this category of patients: additional booster vaccination (stimulation of the humoral response), early detection (screening) of SARS-COV2 for virus-neutralizing antibodies therapy, timely hospitalization of patients.

P1612: ACUTE MYELOID LEUKEMIA IN ADULT PATIENTS: FEATURES OF TREATMENT IN CONTEXT OF THE COVID-19 PANDEMIC.

Background: Acute myeloid leukemia (AML) is a disease with a poor prognosis and a high mortality rate, while patients with acute myeloid leukemia are at increased risk of complications from COVID-19 infection. Aims: To investigate the treatment effectiveness and survival of patients with AML who had the COVID-19 diagnosis confirmed, as well as evaluating predictors of hospital mortality among this group of patients. Methods: From April 2020 to November 2021, Moscow City Clinical Hospital 52 treated 149 patients with AML who had the COVID-19 diagnosis confirmed, with a median age of 61 years (range 18-91), while 56% of patients were over 60 years old. An equal ratio of men and women was performed, 70 and 79 respectively. 13% were diagnosed with APL, while the remaining 87% had other variants of AML. At the time of hospitalization the most of the patients admitted had first- or second-degree pulmonary lesions (28 and 35%, respectively), while 31% of patients performed more severe lesions and the other 6% had no pulmonary damage detected, but they had a positive PCR. At the time of hospitalization, 10 out of 149 patients (7%) achieved complete remissions (CR) and were withdrawn from therapy; 41 patients (28%) were diagnosed with AML for the first time in our clinic; and 98 patients (65%) were already in process of AML treatment in other clinics. The effectiveness of treatment was assessed by the frequency of CR, as well as by the amount of cases of lethality and in-hospital survival rates. The analysis of treatment results was performed as at 01 November 2021. Results: 60 patients (38%) with AML (excluding APL) and confirmed COVID19 received anti-tumor treatment. Palliative care was received by 23 patients (38%), these were predominantly elderly patients with a median age of 71 years (range 45-91 years), with severe COVID-19. 19 patients (32%) with a median age of 38 years (range 21-61 years) received the 7 + 3 treatment protocol. While 18 elderly patients (30%) with a median age of 70 years (range 60-85 years) received the low intensity therapy (low doses of cytarabine, hypomethylation therapy and venetoclax containing regimens). As a result of COVID-19, 79 patients (54%) died. The main factor determining the high mortality rate in our study was the absence of AML CR. The Survival of patients in CR was 96.3±3.6%, which means almost all patients have recovered from COVID-19 and were discharged to continue their treatment in other medical institutions. For patients out of remission and primary patients with AML the survival rate was 38.0±6.5% and 33.7±8.7%, respectively. For the purposes of analyzes of predictors of survival, we divided our patients into 2 groups, due to a great variety of prognosis. Remission status essentially outweighs all other factors, including presence of significant comorbidities, AML variants, isolated cytopenias, severity of COVID-19 (CT, CRP, D-dimer, etc.), nature of AML therapy and other factors. In the group of patients out of AML remission the negative predictive value for hospital survival was shown by age over 60 years, as well as by agranulocytosis, CT 3-4, the use of GCS and transfer to the ICU. Summary/Conclusion: The diagnosis of COVID-19 in patients with AML greatly worsens their life prognosis. The leading predictor of mortality becomes the lack of remission of AML. Thus, antitumor treatment allows achieving CR, which in turn reduces the risk of developing a severe course of coronavirus infection.

Echocardiographic Characteristics of Cardiogenic Shock Patients with and Without Cardiac Arrest.

Cardiac arrest (CA) is associated with worse outcomes in patients with cardiogenic shock (CS). To better understand the contribution of CA on CS, we evaluated transthoracic echocardiography (TTE) parameters in CS patients with and without CA. We retrospectively identified CS patients with a TTE performed near cardiac intensive care unit admission between 2007 to 2018. We compared TTE measurements of left ventricular (LV) and right ventricular (RV) function in patients with and without CA. The primary outcome was all-cause in-hospital mortality, as determined using multivariable logistic regression. We included 1085 patients, 35% of whom had CA. Median age was 70 years and 37% were females. CA patients had higher severity of illness, more invasive mechanical ventilation and greater vasopressor/inotrope use. In-hospital mortality was 31% and was higher in CA patients (45% vs. 23%, p <0.001). Although LV ejection fraction (LVEF) was similar (35% vs. 37%, p = 0.05), CA patients had lower cardiac index, mitral valve E wave peak velocity, E/A ratio and E/e' ratio. TTE variables that were associated with hospital mortality varied, among patients with CA, these included measures of RV pressure and function and among patients without CA, these included parameters reflecting LV systolic function. Doppler assessments of RV systolic dysfunction were the strongest TTE predictors of hospital mortality in CS patients with CA, unlike CS patients without CA in whom LV systolic function was more important. This emphasizes the importance of RV assessment for mortality risk stratification after CA.

Evaluating High-Dimensional Machine Learning Models to Predict Hospital Mortality Among Older Patients With Cancer.

Older hospitalized cancer patients face high risks of hospital mortality. Improved risk stratification could help identify high-risk patients who may benefit from future interventions, although we lack validated tools to predict in-hospital mortality for patients with cancer. We evaluated the ability of a high-dimensional machine learning prediction model to predict inpatient mortality and compared the performance of this model to existing prediction indices. We identified patients with cancer older than 75 years from the National Emergency Department Sample between 2016 and 2018. We constructed a high-dimensional predictive model called Cancer Frailty Assessment Tool (cFAST), which used an extreme gradient boosting algorithm to predict in-hospital mortality. cFAST model inputs included patient demographic, hospital variables, and diagnosis codes. Model performance was assessed with an area under the curve (AUC) from receiver operating characteristic curves, with an AUC of 1.0 indicating perfect prediction. We compared model performance to existing indices including the Modified 5-Item Frailty Index, Charlson comorbidity index, and Hospital Frailty Risk Score. We identified 2,723,330 weighted emergency department visits among older patients with cancer, of whom 144,653 (5.3%) died in the hospital. Our cFAST model included 240 features and demonstrated an AUC of 0.92. Comparator models including the Modified 5-Item Frailty Index, Charlson comorbidity index, and Hospital Frailty Risk Score achieved AUCs of 0.58, 0.62, and 0.71, respectively. Predictive features of the cFAST model included acute conditions (respiratory failure and shock), chronic conditions (lipidemia and hypertension), patient demographics (age and sex), and cancer and treatment characteristics (metastasis and palliative care). High-dimensional machine learning models enabled accurate prediction of in-hospital mortality among older patients with cancer, outperforming existing prediction indices. These models show promise in identifying patients at risk of severe adverse outcomes, although additional validation and research studying clinical implementation of these tools is needed.

Neutrophil-Albumin Ratio as a Predictor of in-Hospital Mortality in Patients with Cardiogenic Shock

Intoduction: Cardiogenic shock (SK) is the most severe phase of the acute heart failure syndrome. One of the most widely studied inflammatory mediators in cardiogenic shock is neutrophils. Albumin has several functions, including in pressure regulation, plays a role as an antioxidant and anti-inflammatory agent. Several studies have shown the association of albumin levels with mortality in patients with cardiogenic shock.&#x0D; Purpose: This study aimed to evaluate the utilization of neutrophil-albumin ratio (NAR) in predicting in-hospital mortality in patients with cardiogenic shock (CS)&#x0D; Patients and methods: This study was an observational study with cross sectional design conducted at the Department of Cardiovascular, Harapan Kita Cardiovascular Hospital. The data were collected from the patient registry (January 2018 to April 2020). The study participants were all patients with cardiogenic shock admitted to our hospital. The endpoint was in-hospital mortality in CS patients. Predictors of hospital mortality were identified using multivariable logistic regression, followed by receiver operator characteristic (ROC) curve analysis and cut-off value for optimal NAR level.&#x0D; Results: A total of 130 patients hospitalized with CS were enrolled in this study, In-hospital mortality was found in 75 (57,7%) patients, among which 102 (78,5%) were male and 101 (77,7%) patients had acute coronary syndrome. There was a significant positive correlation between NAR levels and in-hospital mortality. The multivariate logistic regression showed that NAR was independently associated with increased risk of in-hospital mortality with odd ratio (OR) of 5,81, 95% confidence interval (CI) 2,303 - 14,692, P &lt;0,001. NAR had a prognostic value in predicting in-hospital mortality of CS based on ROC curve analysis (AUC 0,802), with an optimal NAR cut-off value of 25.&#x0D; Conclusion: NAR is independently associated with in-hospital mortality in patients with CS&#x0D; Keywords: Neutropil-albumin ratio, cardiogenic Shock, mortality predictor

Validation of the Pediatric Sequential Organ Failure Assessment Score and Evaluation of Third International Consensus Definitions for Sepsis and Septic Shock Definitions in the Pediatric Emergency Department

Pediatric sepsis definitions have evolved, and some have proposed using the measure used in adults to quantify organ dysfunction, a Sequential Organ Failure Assessment (SOFA) score of 2 or more in the setting of suspected infection. A pediatric adaptation of SOFA (pSOFA) showed excellent discrimination for mortality in critically ill children but has not been evaluated in an emergency department (ED) population. To delineate test characteristics of the pSOFA score for predicting in-hospital mortality among (1) all patients and (2) patients with suspected infection treated in pediatric EDs. This retrospective cohort study took place from January 1, 2012, to January 31, 2020 in 9 US children's hospitals included in the Pediatric Emergency Care Applied Research Network (PECARN) Registry. The data was analyzed from February 1, 2020, to April 18, 2022. All ED visits for patients younger than 18 years were included. ED pSOFA score was assigned by summing maximum pSOFA organ dysfunction components during ED stay (each 0-4 points). In the subset with suspected infection, visit meeting criteria for sepsis (suspected infection with a pSOFA score of 2 or more) and septic shock (suspected infection with vasoactive infusion and serum lactate level >18.0 mg/dL) were identified. Test characteristics of pSOFA scores of 2 or more during the ED stay for hospital mortality. A total of 3 999 528 (female, 47.3%) ED visits were included. pSOFA scores ranged from 0 to 16, with 126 250 visits (3.2%) having a pSOFA score of 2 or more. pSOFA scores of 2 or more had sensitivity of 0.65 (95% CI, 0.62-0.67) and specificity of 0.97 (95% CI, 0.97-0.97), with negative predictive value of 1.0 (95% CI, 1.00-1.00) in predicting hospital mortality. Of 642 868 patients with suspected infection (16.1%), 42 992 (6.7%) met criteria for sepsis, and 374 (0.1%) met criteria for septic shock. Hospital mortality rates for suspected infection (599 502), sepsis (42 992), and septic shock (374) were 0.0%, 0.9%, and 8.0%, respectively. The pSOFA score had similar discrimination for hospital mortality in all ED visits (area under receiver operating characteristic curve, 0.81; 95% CI, 0.79-0.82) and the subset with suspected infection (area under receiver operating characteristic curve, 0.82; 95% CI, 0.80-0.84). In a large, multicenter study of pediatric ED visits, a pSOFA score of 2 or more was uncommon and associated with increased hospital mortality yet had poor sensitivity as a screening tool for hospital mortality. Conversely, children with a pSOFA score of 2 or less were at very low risk of death, with high specificity and negative predictive value. Among patients with suspected infection, patients with pSOFA-defined septic shock demonstrated the highest mortality.