ChatGPT Predicts In-Hospital All-Cause Mortality for Sepsis: In-Context Learning with the Korean Sepsis Alliance Database.

OBJECTIVES: Sepsis is a leading global cause of mortality, and predicting its outcomes is vital for improving patient care. This study explored the capabilities of ChatGPT, a state-of-the-art natural language processing model, in predicting in-hospital mortality for sepsis patients. METHODS: This study utilized data from the Korean Sepsis Alliance (KSA) database, collected between 2019 and 2021, focusing on adult intensive care unit (ICU) patients and aiming to determine whether ChatGPT could predict all-cause mortality after ICU admission at 7 and 30 days. Structured prompts enabled ChatGPT to engage in in-context learning, with the number of patient examples varying from zero to six. The predictive capabilities of ChatGPT-3.5-turbo and ChatGPT-4 were then compared against a gradient boosting model (GBM) using various performance metrics. RESULTS: From the KSA database, 4,786 patients formed the 7-day mortality prediction dataset, of whom 718 died, and 4,025 patients formed the 30-day dataset, with 1,368 deaths. Age and clinical markers (e.g., Sequential Organ Failure Assessment score and lactic acid levels) showed significant differences between survivors and non-survivors in both datasets. For 7-day mortality predictions, the area under the receiver operating characteristic curve (AUROC) was 0.70-0.83 for GPT-4, 0.51-0.70 for GPT-3.5, and 0.79 for GBM. The AUROC for 30-day mortality was 0.51-0.59 for GPT-4, 0.47-0.57 for GPT-3.5, and 0.76 for GBM. Zero-shot predictions using GPT-4 for mortality from ICU admission to day 30 showed AUROCs from the mid-0.60s to 0.75 for GPT-4 and mainly from 0.47 to 0.63 for GPT-3.5. CONCLUSIONS: GPT-4 demonstrated potential in predicting short-term in-hospital mortality, although its performance varied across different evaluation metrics.

Early detection of low QRS voltage and its association with mortality in patients with sepsis.

Various electrocardiographic changes occur during sepsis, but data on the clinical importance of a low QRS voltage in sepsis are still limited. We aimed to evaluate the association between low QRS voltage identified early in sepsis and mortality in patients with sepsis. Between September 2019 and December 2020, all consecutive adult patients diagnosed with sepsis in the emergency room or general ward at Samsung Medical Center were enrolled. Patients without a 12-lead electrocardiogram recorded within 48 h of recognition of sepsis were excluded. In 432 eligible patients, 12-lead electrocardiogram was recorded within the median of 24 min from the first recognition of sepsis, and low QRS voltage was identified in 115 (26.6%) patients. The low QRS group showed more severe organ dysfunction and had higher levels of N-terminal pro-brain natriuretic peptide. The hospital mortality was significantly higher in the low QRS voltage group than in the normal QRS voltage group (49.6% vs. 28.1%, p < 0.001). Similarly, among the 160 patients who required intensive care unit admission, significantly more patients in the low QRS group died in the intensive care unit (35.9% vs. 18.2%, p = 0.021). Low QRS voltage was associated with increased hospital mortality in patients with sepsis.

Sepsis Alert Systems, Mortality, and Adherence in Emergency Departments: A Systematic Review and Meta-Analysis.

Importance: Early detection and management of sepsis are crucial for patient survival. Emergency departments (EDs) play a key role in sepsis management but face challenges in timely response due to high patient volumes. Sepsis alert systems are proposed to expedite diagnosis and treatment initiation per the Surviving Sepsis Campaign guidelines. Objective: To review and analyze the association of sepsis alert systems in EDs with patient outcomes. Data Sources: A thorough search was conducted in PubMed, EMBASE, Web of Science, and the Cochrane Library from January 1, 2004, to November 19, 2023. Study Selection: Studies that evaluated sepsis alert systems specifically designed for adult ED patients were evaluated. Inclusion criteria focused on peer-reviewed, full-text articles in English that reported on mortality, ICU admissions, hospital stay duration, and sepsis management adherence. Exclusion criteria included studies that lacked a control group or quantitative reports. Data Extraction and Synthesis: The review followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline. Two independent reviewers conducted the data extraction using a standardized form. Any disagreements were resolved through discussion. The data were synthesized using a random-effects model due to the expected heterogeneity among the included studies. Main Outcomes and Measures: Key outcomes included mortality, intensive care unit admissions, hospital stay duration, and adherence to the sepsis bundle. Results: Of 3281 initially identified studies, 22 (0.67%) met inclusion criteria, encompassing 19 580 patients. Sepsis alert systems were associated with reduced mortality risk (risk ratio [RR], 0.81; 95% CI, 0.71 to 0.91) and length of hospital stay (standardized mean difference [SMD], -0.15; 95% CI, -0.20 to -0.11). These systems were also associated with better adherence to sepsis bundle elements, notably in terms of shorter time to fluid administration (SMD, -0.42; 95% CI, -0.52 to -0.32), blood culture (SMD, -0.31; 95% CI, -0.40 to -0.21), antibiotic administration (SMD, -0.34; 95% CI, -0.39 to -0.29), and lactate measurement (SMD, -0.15; 95% CI, -0.22 to -0.08). Electronic alerts were particularly associated with reduced mortality (RR, 0.78; 95% CI, 0.67 to 0.92) and adherence with blood culture guidelines (RR, 1.14; 95% CI, 1.03 to 1.27). Conclusions and Relevance: These findings suggest that sepsis alert systems in EDs were associated with better patient outcomes along with better adherence to sepsis management protocols. These systems hold promise for enhancing ED responses to sepsis, potentially leading to better patient outcomes.

Epinephrine induced lactic acidosis during the management of anaphylactic shock: a case report.

In a case of contrast media-induced anaphylactic shock managed with epinephrine, a 57-year-old male developed lactic acidosis without cardiogenic shock or global hypoperfusion, highlighting epinephrine's potential to trigger lactic acidosis. Despite previous management of similar reactions with antihistamines and corticosteroids, this case required intensive care unit admission and emergency intervention, with lactate levels peaking alarmingly. The rapid resolution of acidosis following epinephrine discontinuation underscores the need for careful monitoring and the consideration of alternative vasopressor strategies in severe anaphylaxis, illustrating the complex relationship between epinephrine's metabolic effects and anaphylaxis-induced tissue hypoperfusion.

Restrictive calcium replacement in septic shock: a multicenter before-after intervention study. / Reposición restrictiva de calcio en pacientes con shock séptico: estudio de intervención antes-después.

OBJECTIVES: To study the impact of a restrictive calcium replacement protocol in comparison with a liberal one in patients with septic shock. MATERIAL AND METHODS: Multicenter retrospective before-after study that estimated the impact of implementing a restrictive calcium replacement protocol in patients with septic shock. Patients admitted to an intensive care unit between May 2019 and April 2021 were assigned to liberal calcium replacement, and those admitted between May 2021 and April 2022 were assigned to a restrictive protocol. The primary outcome measure was 28-day mortality. Patients were matched with propensity scores. RESULTS: A total of 644 patients were included; liberal replacement was used in 453 patients and the restrictive replacement in 191. We paired 553 patients according to propensity scores, 386 in the liberal group and 167 in the restrictive group. Mortality did not differ significantly between the groups at 28 days (35.3% vs 32.3%, respectively; hazard ratio, 0.97; 95% CI, 0.72-1.29) or after resolution of septic shock (81.5% vs 83.8%; hazard ratio, 0.89; 95% CI, 0.73-1.09). Nor did scores on the Sepsis-related Organ Failure Assessment scale differ (2.1 vs 2.6; P = 0.20). CONCLUSION: The implementation of a restrictive calcium replacement protocol in patients with septic shock was not associated with a decrease in 28-day mortality in comparison with use of a liberal protocol. However, we were able to reduce calcium replacement without adverse effects.

OBJETIVO: Investigar el efecto de un protocolo de reposición restrictiva de calcio frente a una estrategia liberal en pacientes con shock séptico. METODO: Estudio multicéntrico, antes-después y retrospectivo que evaluó el efecto de la implementación de un protocolo de reposición restrictiva de calcio en pacientes con shock séptico. Los pacientes que ingresaron en unidades de cuidados intensivos (UCI) entre mayo de 2019 y abril de 2021 se asignaron al grupo con administración liberal, y los que se presentaron entre mayo de 2021 y abril de 2022 ­tras la implementación del protocolo­ al grupo con administración restrictiva. La variable de resultado principal fue la mortalidad a 28 días. Se realizó un emparejamiento por puntuación de propensión. RESULTADOS: Se incluyeron 644 pacientes, 453 en el grupo liberal y 191 en el grupo restrictivo. De los que 553 se emparejaron (386 en el grupo liberal, y 167 en el grupo restrictivo). No hubo diferencias entre los dos grupos en la mortalidad a los 28 días (35,3% vs 32,3%; HR: 0,97; IC 95%: 0,72-1,29), en la finalización del shock (81,5% vs a 83,8%; HR: 0,89; IC 95%: 0,73-1,09) ni en la puntuación de la escala SOFA (2,1 vs 2,6; p = 0,20). CONCLUSIONES: La implementación de un protocolo de administración restrictiva de calcio, en pacientes con shock séptico, no se asoció a una disminución de la mortalidad a los 28 días en comparación con una administración liberal. No obstante, la reposición de calcio podría reducirse sin efectos adversos.

Massive Transfusion Protocol and Outcome of Patients with Acute Variceal Bleeding.

Background/Aims: The massive transfusion protocol (MTP) can improve the outcomes of trauma patients with hemorrhagic shock and some patients with non-traumatic hemorrhagic shock. However, no information is available regarding whether MTP can improve the outcomes of acute variceal bleeding (AVB). This study aimed to determine the effects of MTP on the outcomes of patients with AVB. Methods: Consecutive patients (n = 218) with AVB who did not have current malignancy and visited the emergency room between July 2014 and June 2022 were analyzed. 42-day mortality and failure to control the bleeding were compared between patients with and without MTP activation. Additionally, propensity-score matching was conducted. Results: The amount of blood product transfused was higher in the MTP group. The 42-day mortality rate (42.1% vs. 1.5%, p < 0.001) and the rate of failure to control bleeding (36.8% vs. 0.5%, p < 0.001) were significantly higher in those who received blood transfusions by MTP. MTP was an independent factor associated with 42-day mortality in the multivariable-adjusted analysis (HR 21.05; 95% CI 3.07-144.21, p = 0.002, HR 24.04; 95% CI 3.41-169.31, p = 0.001). The MTP group showed consistently higher 42-day mortality and failure to control bleeding in all subgroup analyses, stratified by systolic blood pressure, hemoglobin level, and the model for end-stage liver disease score. The MTP group also showed higher 42-day mortality (42.9% vs. 0%, p = 0.001) and failure to control bleeding (42.9% vs. 0%, p = 0.001) in a propensity score-matched analysis (n = 52). Conclusions: MTP was associated with poor outcomes in patients with AVB. Further studies are needed to see whether MTP can be an option for patients with massive AVB.

Overt disseminated intravascular coagulation and antithrombin III predict bleeding and in-hospital mortality in patients undergoing extracorporeal membrane oxygenation.

Background: Limited data are available on the relationship of disseminated intravascular coagulation (DIC) with mortality in patients receiving extracorporeal membrane oxygenation (ECMO). Thus, we investigated the association of DIC score and antithrombin (AT) III with clinical outcomes in patients undergoing ECMO. Methods: We analyzed 703 patients who underwent ECMO between January 2014 and May 2022 at Samsung Medical Center. The DIC score was calculated using laboratory findings within 24 h of the ECMO initiation, and ≥ 5 was defined as overt DIC. In addition, the AT III level was measured to identify the correlation with the DIC score. Results: Among the study patients, 169 (24.0%) were diagnosed with overt DIC (DIC group) during early maintenance therapy. In-hospital mortality was significantly higher in the DIC group than in the non-DIC group (55.0% vs. 36.5%, p < 0.001). Bleeding events were significantly higher in the group of patients with a DIC score of 7 or 8 than in the other group (20.8% vs. 8.4%, p = 0.038). DIC score negatively correlated with AT III level (r = -0.417, p < 0.001). The predictive performance of AT III for overt DIC had statistical significance with a c-static of 0.81 (95% confidence interval (CI), 0.77-0.84, p < 0.001). Conclusion: Overt DIC was associated with higher in-hospital mortality and a tendency to bleed in ECMO patients. Furthermore, AT III plasma levels can easily predict overt DIC in patients undergoing ECMO. These findings suggest that monitoring AT III plasma levels may be important in the management of ECMO.

Shock indices are associated with in-hospital mortality among patients with septic shock and normal left ventricular ejection fraction.

BACKGROUND: The understanding of shock indices in patients with septic shock is limited, and their values may vary depending on cardiac function. METHODS: This prospective cohort study was conducted across 20 university-affiliated hospitals (21 intensive care units [ICUs]). Adult patients (≥19 years) with septic shock admitted to the ICUs during a 29-month period were included. The shock index (SI), diastolic shock index (DSI), modified shock index (MSI), and age shock index (Age-SI) were calculated at sepsis recognition (time zero) and ICU admission. Left ventricular (LV) function was categorized as either normal LV ejection fraction (LVEF ≥ 50%) or decreased LVEF (<50%). RESULTS: Among the 1,194 patients with septic shock, 392 (32.8%) who underwent echocardiography within 24 h of time zero were included in the final analysis (normal LVEF: n = 246; decreased LVEF: n = 146). In patients with normal LVEF, only survivors demonstrated significant improvement in SI, DSI, MSI, and Age-SI values from time zero to ICU admission; however, no notable improvements were found in all patients with decreased LVEF. The completion of vasopressor or fluid bundle components was significantly associated with improved indices in patients with normal LVEF, but not in those with decreased LVEF. In multivariable analysis, each of the four indices at ICU admission was significantly associated with in-hospital mortality (P < 0.05) among patients with normal LVEF; however, discrimination power was better in the indices for patients with lower lactate levels (≤ 4.0 mmol/L), compared to those with higher lactate levels. CONCLUSIONS: The SI, DSI, MSI, and Age-SI at ICU admission were significantly associated with in-hospital mortality in patients with septic shock and normal LVEF, which was not found in those with decreased LVEF. Our study emphasizes the importance of interpreting shock indices in the context of LV function in septic shock.

Ambulatory Status at Discharge Predicts Six-Month Mortality in Patients with COVID-19: A Retrospective Cohort Study.

This retrospective cohort study aimed to evaluate the association between ambulatory status at discharge and six-month post-discharge mortality among adults with coronavirus disease (COVID-19). We analyzed data from 398 patients aged over 18 admitted to a tertiary hospital in South Korea between December 2019 and June 2022. Patients were classified into two groups based on their ambulatory status at discharge: ambulatory (able to walk independently, n = 286) and non-ambulatory (unable to walk independently, requiring wheelchair or bed-bound, n = 112). Our analysis revealed that six-month survival rates were significantly higher in the ambulatory group (94.2%) compared to the non-ambulatory group (84.4%). Multivariate analysis identified ambulatory status at discharge (p = 0.047) and pre-existing malignancy (p = 0.007) as significant prognostic factors for post-discharge survival. This study highlights that the ability to walk independently at discharge is a crucial predictor of six-month survival in COVID-19 patients. These findings emphasize the need for interventions to improve the physical performance of non-ambulatory patients, potentially enhancing their survival prospects. This underscores the importance of targeted rehabilitation and physical therapy for the comprehensive care of COVID-19 survivors.

Comparison of noninvasive cardiac output and stroke volume measurements using electrical impedance tomography with invasive methods in a swine model.

Pulmonary artery catheterization (PAC) has been used as a clinical standard for cardiac output (CO) measurements on humans. On animals, however, an ultrasonic flow sensor (UFS) placed around the ascending aorta or pulmonary artery can measure CO and stroke volume (SV) more accurately. The objective of this paper is to compare CO and SV measurements using a noninvasive electrical impedance tomography (EIT) device and three invasive devices using UFS, PAC-CCO (continuous CO) and arterial pressure-based CO (APCO). Thirty-two pigs were anesthetized and mechanically ventilated. A UFS was placed around the pulmonary artery through thoracotomy in 11 of them, while the EIT, PAC-CCO and APCO devices were used on all of them. Afterload and contractility were changed pharmacologically, while preload was changed through bleeding and injection of fluid or blood. Twenty-three pigs completed the experiment. Among 23, the UFS was used on 7 pigs around the pulmonary artery. The percentage error (PE) between COUFS and COEIT was 26.1%, and the 10-min concordance was 92.5%. Between SVUFS and SVEIT, the PE was 24.8%, and the 10-min concordance was 94.2%. On analyzing the data from all 23 pigs, the PE between time-delay-adjusted COPAC-CCO and COEIT was 34.6%, and the 10-min concordance was 81.1%. Our results suggest that the performance of the EIT device in measuring dynamic changes of CO and SV on mechanically-ventilated pigs under different cardiac preload, afterload and contractility conditions is at least comparable to that of the PAC-CCO device. Clinical studies are needed to evaluate the utility of the EIT device as a noninvasive hemodynamic monitoring tool.

The association of the Sequential Organ Failure Assessment score at intensive care unit discharge with intensive care unit readmission in the cardiac intensive care unit.

AIMS: Unplanned intensive care unit (ICU) readmissions contribute to increased morbidity, mortality, and healthcare costs. The severity of patient illness at ICU discharge may predict early ICU readmission. Thus, in this study, we investigated the association of cardiac ICU (CICU) discharge Sequential Organ Failure Assessment (SOFA) score with unplanned CICU readmission in patients admitted to the CICU. METHODS AND RESULTS: We retrospectively reviewed the hospital medical records of 4659 patients who were admitted to the CICU from 2012 to 18. Sequential Organ Failure Assessment scores at CICU admission and discharge were obtained. The predictive performance of organ failure scoring was evaluated by using area under the receiver operating characteristic (AUROC) curves. The primary outcome was unplanned CICU readmission. Of the 3949 patients successfully discharged from the CICU, 184 (4.7%) had an unplanned CICU readmission or they experienced a deteriorated condition but died without being readmitted to the CICU (readmission group). The readmission group had significantly higher rates of organ failure in all organ systems at both CICU admission and discharge than the non-readmission group. The AUROC of the discharge SOFA score for CICU readmission was 0.731, showing good predictive performance. The AUROC of the discharge SOFA score was significantly greater than that of either the initial SOFA score (P = 0.020) or the Acute Physiology and Chronic Health Evaluation II score (P < 0.001). In the multivariable regression analysis, SOFA score, overweight or obese status, history of heart failure, and acute heart failure as reasons for ICU admission were independent predictors of unplanned ICU readmission during the same hospital stay. CONCLUSION: The discharge SOFA score may identify patients at a higher risk of unplanned CICU readmission, enabling targeted interventions to reduce readmission rates and improve patient outcomes.

Machine learning methods for developing a predictive model of the incidence of delirium in cardiac intensive care units.

INTRODUCTION AND OBJECTIVES: Delirium, recognized as a crucial prognostic factor in the cardiac intensive care unit (CICU), has evolved in response to the changing demographics among critically ill cardiac patients. This study aimed to create a predictive model for delirium for patients in the CICU. METHODS: This study included consecutive patients admitted to the CICU of the Samsung Medical Center. To assess the candidate variables for the model: we applied the following machine learning methods: random forest, extreme gradient boosting, partial least squares, and Plmnet-elastic.net. After selecting relevant variables, we performed a logistic regression analysis to derive the model formula. Internal validation was conducted using 100-repeated hold-out validation. RESULTS: We analyzed 2774 patients, 677 (24.4%) of whom developed delirium in the CICU. Machine learning-based models showed good predictive performance. Clinically significant and frequently important predictors were selected to construct a delirium prediction scoring model for CICU patients. The model included albumin level, international normalized ratio, blood urea nitrogen, white blood cell count, C-reactive protein level, age, heart rate, and mechanical ventilation. The model had an area under the receiver operating characteristics curve (AUROC) of 0.861 (95%CI, 0.843-0.879). Similar results were obtained in internal validation with 100-repeated cross-validation (AUROC, 0.854; 95%CI, 0.826-0.883). CONCLUSIONS: Using variables frequently ranked as highly important in four machine learning methods, we created a novel delirium prediction model. This model could serve as a useful and simple tool for risk stratification for the occurrence of delirium at the patient's bedside in the CICU.

Clinical effects of bacteremia in sepsis patients with community-acquired pneumonia.

BACKGROUND: Data regarding the clinical effects of bacteremia on severe community-acquired pneumonia (CAP) are limited. Thus, we investigated clinical characteristics and outcomes of severe CAP patients with bacteremia compared with those of subjects without bacteremia. In addition, we evaluated clinical factors associated with bacteremia at the time of sepsis awareness. METHODS: We enrolled sepsis patients diagnosed with CAP at emergency departments (EDs) from an ongoing nationwide multicenter observational registry, the Korean Sepsis Alliance, between September 2019 and December 2020. For evaluation of clinical factors associated with bacteremia, we divided eligible patients into bacteremia and non-bacteremia groups, and logistic regression analysis was performed using the clinical characteristics at the time of sepsis awareness. RESULT: During the study period, 1,510 (47.9%) sepsis patients were caused by CAP, and bacteremia was identified in 212 (14.0%) patients. Septic shock occurred more frequently in the bacteremia group than in the non-bacteremia group (27.4% vs. 14.8%; p < 0.001). In multivariable analysis, hematologic malignancies and septic shock were associated with an increased risk of bacteremia. However, chronic lung disease was associated with a decreased risk of bacteremia. Hospital mortality was significantly higher in the bacteremia group than in the non-bacteremia group (27.3% vs. 40.6%, p < 0.001). The most prevalent pathogen in blood culture was Klebsiella pneumoniae followed by Escherichia coli in gram-negative pathogens. CONCLUSION: The incidence of bacteremia in severe CAP was low at 14.0%, but the occurrence of bacteremia was associated with increased hospital mortality. In severe CAP, hematologic malignancies and septic shock were associated with an increased risk of bacteremia.