Journal of Intensive Care Medicine最新文献

Background: Acute Kidney Injury (AKI), a leading organ failure cause in critical patients, demands early high-risk identification to enhance outcomes. Yet comparative analyses of diagnostic and prognostic machine learning (ML) models across multiple post-admission timeframes are lacking.

Methods: Using MIMIC-IV, we carried out using the Boruta algorithm for feature selection, developing and comparing six ML models to predict AKI risk at 0-24, 24-48, 48-72, 0-48, and 0-72 h post-ICU admission. Model performance was evaluated using the Area Under the Curve (AUC) and confusion matrix. Decision Curve and calibration analyses assessed clinical applicability. We compared models with Sequential Organ Failure Assessment (SOFA) and SAPSII scores to evaluate the accuracy of the ML models. Finally, Shapley Additive Explanations (SHAP) values interpreted and visualized key features of the optimal model.

Results: Our study involved 2092 trauma Intensive Care Unit (ICU) patients. Using the 17 selected out of the 48 features among trauma patients 24 h after ICU admissions, among the six ML models and two scoring systems, all ML models outperformed SOFA and SAPS II, and the extreme gradient boosting (XGBoost) exhibited the best performance, achieving an AUC of 0.948 (95% CI [0.929-0.966]) for AKI prediction within 24 h of admission, with an AUC of 0.941 ([0.892-0.917]) and 0.878 ([0.863-0.892]) at 0-48 and 0-72 h period, respectively. However, their predictive accuracies were very limited at 24-48 h (AUC 0.602 [0.562-0.643]) and 48-72 h (AUC 0.490 [0.429-0.551]), respectively. Urine output per kilogram per hour at 6 and 12 h and age were the most important features identified through SHAP analysis.

Conclusions: Our study found ML models excel in diagnosing AKI risk in ICU trauma patients but have limited prognostic accuracy at 24-48 and 48-72 h post-admission. Further research is needed to improve this using time-series ML models with optimal windows.

Pulmonary hypertension, characterized by elevated pressures in the pulmonary arteries leading to abnormalities in right ventricular function, may lead to competing demands between the pulmonary and systemic circulation during sepsis and septic shock. As a result, management of pulmonary hypertension in sepsis, including identifying the source of infection, maintaining hemodynamic stability and continuing or transitioning pulmonary hypertension-specific therapies can often be challenging. The goal of this review is to highlight factors to consider in the evaluation and management of patients with pulmonary hypertension and sepsis.

AimTo evaluate the efficiency of combined diaphragm and intercostal muscle ultrasound assessment in predicting the extubation outcome in mechanically ventilated patients with sepsis.MethodsThis study was a prospective observational study of septic patients consecutively admitted to the hospital from October 2022 to October 2023 for mechanical ventilation. During the period when the patients passed the ventilator weaning screening and spontaneous breathing trial (SBT), ultrasound evaluation of the diaphragm and intercostal muscles was performed to measure diaphragm excursion (DE), diaphragm thickening fraction (TFD) and intercostal muscle thickening fraction (TFic). The patients were divided into the successful extubation group (89 cases) and the failed extubation group (15 cases) according to the extubation results. ROC curves were used to analyze the effects of diaphragm ultrasound and intercostal muscle ultrasound alone and in combination to predict extubation outcomes.ResultsTFic and TFic/TFD values were significantly higher in the failed extubation group than in the successful extubation group during extubation (P < 0.05). The area under the ROC curve (AUROC) of DE, TFD, and TFic to predict extubation failure in mechanically ventilated patients with sepsis before extubation were 0.689, 0.657, and 0.769, respectively, whereas the combined indexes, such as TFic/TFD and TFic &TFD_mix had AUROCs of 0.867 and 0.860, respectively. TFic/TFD with a cutoff value of >0.95, had a sensitivity of 86.7% and specificity of 75.3% in predicting extubation failure, and TFic &TFD_mix with a cutoff value of >0.13, had a sensitivity of 86.6% and specificity of 80.9% in predicting extubation failure. Conclusion: The combination of diaphragm and intercostal muscle ultrasound assessment might effectively predict the extubation outcome in mechanically ventilated patients with sepsis.

Background: Sepsis management in elderly populations presents unique challenges due to age-related physiological changes and comorbidities. Current guidelines remain conflicted regarding optimal antibiotic timing. We conducted a retrospective, multicenter study to evaluate the association between antibiotic administration timing and short-term and long-term outcomes in elderly sepsis patients.

Methods: This retrospective cohort study analyzed data from the MIMIC-IV (v3.1) database. Patients were categorized into the early group (antibiotics initiated within 1 h) and the late group (antibiotics initiated >1 h after diagnosis). Further analyses were stratified by shock status (septic shock vs non-septic shock) and pathogen type (Gram-positive vs Gram-negative bacteria). Multivariable Cox regression assessed associations between antibiotic administration timing and 28-/180-/365-day hospital mortality. Restricted cubic spline models evaluated dose-response relationships. The primary outcome was 28-day hospital mortality. Secondary outcomes included 180-day and 365-day mortality rates, along with the incidence of continuous renal replacement therapy (CRRT) and mechanical ventilation requirements.

Results: A total of 12,425 patients met the inclusion criteria from the MIMIC-IV database. The multivariable-adjusted analysis demonstrated that delayed antibiotics administration was significantly associated with a 35% increased risk of 28-day all-cause hospital mortality (HR = 1.35, 95% CI 1.22-1.52; P < 0.001), a 43% elevated 180-day hospital mortality risk (HR = 1.43, 95% CI 1.30-1.56; P < 0.001), and a 45% higher 365-day mortality risk (HR = 1.45, 95% CI 1.33-1.56; P < 0.001). Stratified analyses revealed mortality benefits persisted in non-shock patients (28-day HR = 1.31, P < 0.001) and Gram-positive infections (28-day HR = 1.63, P < 0.001), whereas no significant associations emerged in septic shock (28-day HR = 0.82, 95%CI 0.65-1.03; P = 0.081) or Gram-negative infections (HR = 1.04, 95%CI 0.87-1.24; P = 0.692). A linear relationship was observed between antibiotic delay and mortality (Nonlinear P = 0.88).

Conclusions: Early antibiotic administration improves survival in elderly sepsis patients, particularly non-shock cases and Gram-positive infections. These insights advocate the importance of individualized selection based on patients' clinical context in critical care practice.

Sepsis associated encephalopathy (SAE) is common in patients with sepsis, and the occurrence of SAE often indicated adverse outcomes. In recent years, the imbalance of the intestinal microbiota and metabolites have been found to be related to the occurrence of SAE, and this regulation is often accompanied by the activation of the immune system. Possible mechanism still needs to be clarified. Intestinal flora disturbances and altered metabolic profiles are often accompanied by changes in the levels of small molecule metabolites, some of which are critical for the maintenance of brain functional homeostasis, such as short-chain fatty acids (SCFAs). These changes further affect the permeability of the blood-brain barrier and the activation of the central and peripheral immune system, and finally promote the release of inflammatory cytokines and the activation of immune cells. Targeting intestinal microbiota profile, small molecule metabolite, or neurostimulation regulation may be potential therapeutic methods for SAE, such as amino acid supplements, microbiota transplantation, or other metabolite level regulation drugs. Our review will summarize the intestinal flora disturbances, metabolic profiles, neuro-immunoinflammatory changes and related possible drug intervention. These findings may provide the possibility for further exploration of the mechanisms and treatment methods of SAE.

BackgroundArterial catheterization serves as a cornerstone monitoring modality in septic shock management, enabling continuous hemodynamic assessment and serial blood gas analysis. Despite its widespread use, mortality benefits and optimal timing for catheter insertion remains undefined.MethodsAn analysis of clinical data from 6,485 critically ill adult patients, identified as meeting the Sepsis-3 criteria for septic shock, was conducted utilizing the MIMIC-IV database. Through entropy-balanced propensity score matching (PSM, 1:1 ratio) and doubly robust estimation with inverse probability weighting, we compared outcomes between catheterized (≤24 h post-admission) and non-catheterized groups. Restricted cubic spline (RCS) modeling characterized nonlinear temporal associations. The evaluation encompassed both primary and secondary endpoints, including 28-day mortality, mortality within the ICU and hospital settings, length of stay, CRRT requirements, and physiological resuscitation metrics.ResultsAfter PSM (1,416 patients from initial 6,485) with 1:1 ratio, arterial catheterization exhibited significantly reduced mortality across all measured outcomes compared to non-catheterized controls: 28-day mortality (26.1% vs 43.9%; aHR 0.62, 95%CI 0.51-0.75), ICU mortality (aHR 0.76, 0.61-0.94), and in-hospital mortality (HR 0.70, 0.58-0.86), all P < .05. Arterial catheterization was associated with a shorten ICU stay by 0.52 days (95%CI 0.18-0.82, P = .002) and improved physiological parameters. Restricted cubic splines identified optimal intervention timing at 204-290 min post-admission through U-shaped mortality risk association.ConclusionIn a cohort of critically ill patients with septic shock, early peripheral arterial catheterization is significantly associated with improved 28-day mortality outcomes.

PurposeThis study aimed to investigate in-patient mortality and predictors of death associated with convulsive status epilepticus (CSE) in a large nationwide cohort and create a simplified predictive score for in-hospital mortality.MethodsRetrospective data from the National Inpatient Sample (NIS) database between 2007 and 2014 were analyzed, including 123,082 adults with CSE. Univariate logistic testing identified admission variables, neurological and medical complications associated with mortality. A simplified clinical prediction score, called M³A²S²H, was generated using variables that were frequent (>1%) and had a significant impact on mortality.ResultsThe overall hospital mortality rate was 3.5%. Univariate analysis revealed that older age, female gender, past medical history, and acute hospital conditions were related to mortality. After reclassification, a final multivariable model with 27 clinical variables was constructed, and the eight strongest predictors were included in the M³A²S²H score: hypoxic-ischemic encephalopathy/cardiac arrest (2 points); age >60 years, acute symptomatic CSE, invasive mechanical ventilation, sepsis, metastases, and chronic liver failure (all 1 point); and medication nonadherence (-1 point). The mortality rate among patients with ≤0, 1, 2, 3, 4, or ≥5 of these risk factors progressively increased from 0.2%, 2.1%, 7.8%, 20.3%, 31.9%, to 50.0% (P < 0.0001). Additionally, a similar stepwise trend was observed regarding discharge to a facility versus home without services (P < 0.0001).ConclusionsThis study demonstrates that mortality in CSE cases occurs in 3.5% of adult hospital admissions. Identification of specific acute and chronic conditions using the standardized M³A²S²H score can help predict the risk of death or disability even in hospitals without advanced brain monitoring.

Introduction: Tracheal intubation (TI) of an obstetrical patient around the time of delivery can be an upsetting event for involved providers. It can also cause an unpredictable use of intensive care resources. Its incidence is currently poorly characterized in the literature. We analyzed the 2019 National Inpatient Sample (NIS) to assess the incidence rate and associated risk factors. Methods: Patients were identified by International Classification of Diseases, 10th edition codes for delivery of a child. Measured endpoints were the incidence of TI and factors associated. Categorical variables were compared using Chi squared or Fisher's Exact. Continuous variables were compared using the Student T-test or the Mann Whitney rank sum U-test. A logistic regression model was created to determine the odds for each variable contributing to TI. A P value of 0.05 was considered the minimum standard for significance. Results: There was a low rate of TI (0.03%). Mortality was rare (0.004%) and there was a higher rate of mortality among patients who underwent tracheal intubation (5.5% vs 0.003% among patients not intubated, P < .001). The majority of intubations occurred among patients who delivered via CD. Pneumonia, cardiomyopathy, eclampsia, and postpartum hemorrhage were all independently associated with increased odds for TI. Conclusions: There are risk factors which may increase the likelihood for tracheal intubation. The diagnosis of a cardiomyopathy was strongly associated with an increased odds for TI and may result from acute respiratory failure. PPH and eclampsia were also associated with a greater odds for intubation.

PurposeAppropriateness of care is a fundamental yet often underexamined aspect of critical care medicine. Drawing on Kant's notion of appropriateness as providing "no more and no less than the matter requires," we explore how clinicians can align medical indication with patient or surrogate consent. Misalignments in this regard may result in overtreatment or undertreatment, contributing to moral distress, burnout, and dissatisfaction among families and clinicians.MethodsWe present a case report of an elderly patient whose clinical course in the intensive care unit (ICU) traverses all four quadrants of a proposed framework-the Fourfold Table of Appropriateness. This conceptual tool classifies medical interventions according to two dimensions: medical indication (yes/no) and patient or surrogate consent (yes/no). Each quadrant is associated with ethical and clinical implications and is color-coded like a traffic light to support real-time decision-making: green (appropriate), yellow (caution), red (stop).ResultsThe case illustrates how alignment between indication and consent fosters goal-concordant care. It also demonstrates how therapeutic obstinance, paternalism, or emotionally driven demands may lead to inappropriate care. Each episode reveals different challenges-physician biases, surrogate distress, unclear prognoses-and highlights the role of communication, shared decision-making, and clinical humility.ConclusionThe Fourfold Table provides a didactic framework to guide structured reflection among ICU clinicians. By naming and examining patterns of inappropriateness, it fosters ethical awareness and better communication practices. This case-based approach highlights the necessity of early recognition of misalignments, the importance of addressing emotional and cognitive biases, and the value of interventions such as time-limited trials and family-centered care. Ultimately, this framework contributes to more compassionate, appropriate, and goal-aligned critical care.

BackgroundDelirium is a common complication of illness. Patients who experience delirium are at risk for worse outcomes during and after hospitalization. This study aims to describe rates of, and factors associated with, delirium documentation during care transitions for patients who screened positive for delirium in the pediatric intensive care unit (PICU) at transfer to the pediatric hospital medicine (PHM) service and discharge.MethodsDemographic and clinical characteristics were collected retrospectively on patients ages 0-21 years, without developmental delay, who screened positive via Cornell Assessment of Pediatric Delirium (CAPD) before transferring from PICU to PHM service of a tertiary-care children's hospital from 2016-2022. Primary outcomes were documentation of "delirium" at PICU transfer and hospital discharge. Statistical analysis included bivariate analysis and multivariate logistic regression.ResultsOf 337 encounters, 66 transfer (20%) and 62 discharge notes (18%) documented delirium. On bivariate analysis, older age, female sex, Hispanic ethnicity, prolonged and elevated CAPD scoring, longer PICU and hospital length of stay, mechanical ventilation (MV), and psychiatry consultation were associated with documentation at transfer. On logistic regression, Black race decreased odds (OR 0.275, 95% CI 0.08-0.84) while psychiatry consultation (OR 66.82, 24.45-212.25) and invasive MV (OR 6.495, 2.13-22.34) increased odds of documentation. Discharge documentation demonstrated similar associations, except sex and ethnicity were not associated, while neurology consultation was positively associated with documentation. On logistic regression, psychiatry consultation (OR 36.01, 14.51-100.71) and invasive MV (OR 2.96, 1.09-8.83) increased odds of documentation at discharge.ConclusionsDespite a validated screening tool, pediatric providers often fail to document delirium at PICU transfer and hospital discharge. Lack of documentation may leave patients and families at risk for worse outcomes.