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.

Background: Limited data are available on the mortality rates of patients receiving extracorporeal membrane oxygenation (ECMO) support for coronavirus disease 2019 (COVID-19). We aimed to analyze the relationship between COVID-19 and clinical outcomes for patients receiving ECMO. Methods: We retrospectively investigated patients with COVID-19 pneumonia requiring ECMO in 19 hospitals across Korea from January 1, 2020 to August 31, 2021. The primary outcome was the 90-day mortality after ECMO initiation. We performed multivariate analysis using a logistic regression model to estimate the odds ratio (OR) of 90-day mortality. Survival differences were analyzed using the Kaplan–Meier (KM) method. Results: Of 127 patients with COVID-19 pneumonia who received ECMO, 70 patients (55.1%) died within 90 days of ECMO initiation. The median age was 64 years, and 63% of patients were male. The incidence of ECMO was increased with age but was decreased after 70 years of age. However, the survival rate was decreased linearly with age. In multivariate analysis, age (OR, 1.048; 95% confidence interval [CI], 1.010–1.089; P = 0.014) and receipt of continuous renal replacement therapy (CRRT) (OR, 3.069; 95% CI, 1.312–7.180; P = 0.010) were significantly associated with an increased risk of 90-day mortality. KM curves showed significant differences in survival between groups according to age (65 years) (log-rank P = 0.021) and receipt of CRRT (log-rank P = 0.004). Conclusion Older age and receipt of CRRT were associated with higher mortality rates among patients with COVID-19 who received ECMO.

Purpose: Real-world experience with tocilizumab in combination with dexamethasone in patients with severe coronavirus disease (COVID-19) needs to be investigated. Materials and Methods: A retrospective cohort study was conducted to evaluate the effect of severity-adjusted dosing of dexamethasone in combination with tocilizumab for severe COVID-19 from August 2020 to August 2021. The primary endpoint was 30-day clinical recovery, which was defined as no oxygen requirement or referral after recovery. Results: A total of 66 patients were evaluated, including 33 patients in the dexamethasone (Dexa) group and 33 patients in the dexamethasone plus tocilizumab (DexaToci) group. The DexaToci group showed a statistically significant benefit in 30-day clinical recovery, compared to the Dexa group (p=0.024). In multivariable analyses, peak FiO2 within 3 days and tocilizumab combination were consistently significant for 30-day recovery (all p<0.05). The DexaToci group showed a significantly steeper decrease in FiO2 (-4.2±2.6) than the Dexa group (−2.7±2.6; p=0.021) by hospital day 15. The duration of oxygen requirement was significantly shorter in the DexaToci group than the Dexa group (median, 10.0 days vs. 17.0 days; p=0.006). Infectious complications and cellular and humoral immune responses against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in the convalescence stage were not different between the two groups. Conclusion A combination of severity-adjusted dexamethasone and tocilizumab for the treatment of severe COVID-19 improved clinical recovery without increasing infectious complications or hindering the immune response against SARS-CoV-2.

Purpose: In the intensive care unit, a peripherally inserted central catheter (PICC) may be an alternative option to standard central venous catheters, particularly in patients with coagulopathies or at high risk of infection. The purpose of this research was to assess the feasibility of bedside ultrasound (US)-guided PICC placement by critical care fellows on intensive care units.
Methods: All bedside US-PICCs inserted by critical care fellows from July 2013 to September 2015 were retrospectively reviewed focusing on the rate of successful insertion, complications of insertion, or during maintenance.
Results: A total of 177 US-guided PICCs were inserted in 163 patients and included in the analysis. The median age was 62 years (IQR 50-70 years) and 104 cases (58.8%) were male. There were 172 cases (90.4%) of PICCs inserted in the upper arm. Anticoagulant therapy was used in 26 patients (14.7%) and 8 patients (5.2%) had severe coagulopathies. The median procedural time was 30 minutes (IQR 19-45 minutes). Insertion success rate was 93.2%, and there were no major complications during insertions except for malposition (12.1%). Catheters remained in place for a total of 3,878 days (median 16 days: IQR 8-31 days). There was only 1 case (0.6%) of catheter-related bloodstream infection, and 2 cases (1.2%) of symptomatic venous thromboembolism.
Conclusion Bedside US-guided placement of PICCs by critical care fellows is safe and feasible. The success rate of the procedure was “acceptable,” and was not associated with significant risks of infectious and non-infectious complications, even in patients with coagulopathies.

Purpose: In the intensive care unit, a peripherally inserted central catheter (PICC) may be an alternative option to standard central venous catheters, particularly in patients with coagulopathies or at high risk of infection. The purpose of this research was to assess the feasibility of bedside ultrasound (US)-guided PICC placement by critical care fellows on intensive care units.
Methods: All bedside US-PICCs inserted by critical care fellows from July 2013 to September 2015 were retrospectively reviewed focusing on the rate of successful insertion, complications of insertion, or during maintenance.
Results: A total of 177 US-guided PICCs were inserted in 163 patients and included in the analysis. The median age was 62 years (IQR 50-70 years) and 104 cases (58.8%) were male. There were 172 cases (90.4%) of PICCs inserted in the upper arm. Anticoagulant therapy was used in 26 patients (14.7%) and 8 patients (5.2%) had severe coagulopathies. The median procedural time was 30 minutes (IQR 19-45 minutes). Insertion success rate was 93.2%, and there were no major complications during insertions except for malposition (12.1%). Catheters remained in place for a total of 3,878 days (median 16 days: IQR 8-31 days). There was only 1 case (0.6%) of catheter-related bloodstream infection, and 2 cases (1.2%) of symptomatic venous thromboembolism.
Conclusion Bedside US-guided placement of PICCs by critical care fellows is safe and feasible. The success rate of the procedure was “acceptable,” and was not associated with significant risks of infectious and non-infectious complications, even in patients with coagulopathies.

Purpose: The aim of this study was to analyze the temporal change of resuscitation fluid use based on all fluids administered in a surgical intensive care unit (ICU). Methods: The administration of resuscitation fluid to all patients admitted to a surgical ICU of a tertiary referral hospital was investigated from 2008 to 2015. The types and volumes of fluid, and laboratory data taken within 7 days after ICU admission were evaluated. Resuscitation fluids were defined as fluids infused according to stat orders, rather than routine orders. Results: There were a total of 8,885 admissions to the ICU for 7,886 patients. The volumetric proportion of crystalloid to total resuscitation fluids increased significantly over the study period (p < 0.001; 79.6% in 2008; 93.7% in 2015). Although the proportion of 0.9% saline to crystalloids decreased, that of balanced solutions increased (p < 0.001; 29.5% in 2008; 55.6% in 2015). The use of colloids decreased from 20.4% in 2008, to 6.3% in 2015 (p < 0.001). Proportions calculated using the number of individual fluids administered revealed trends similar to those calculated using volumetric data. The amount of infused 0.9% saline was weakly correlated with the lowest blood pH and the highest serum chloride levels (ρ = -0.26 and 0.19, respectively). Conclusion Changes in the trends of fluid resuscitation practice were noted in a single surgical ICU over the 8-year study period. Crystalloid use increased owing to a rise in the utilization of balanced solutions with a downward trend in colloid use.

Purpose: The aim of this study was to analyze the temporal change of resuscitation fluid use based on all fluids administered in a surgical intensive care unit (ICU). Methods: The administration of resuscitation fluid to all patients admitted to a surgical ICU of a tertiary referral hospital was investigated from 2008 to 2015. The types and volumes of fluid, and laboratory data taken within 7 days after ICU admission were evaluated. Resuscitation fluids were defined as fluids infused according to stat orders, rather than routine orders. Results: There were a total of 8,885 admissions to the ICU for 7,886 patients. The volumetric proportion of crystalloid to total resuscitation fluids increased significantly over the study period (p < 0.001; 79.6% in 2008; 93.7% in 2015). Although the proportion of 0.9% saline to crystalloids decreased, that of balanced solutions increased (p < 0.001; 29.5% in 2008; 55.6% in 2015). The use of colloids decreased from 20.4% in 2008, to 6.3% in 2015 (p < 0.001). Proportions calculated using the number of individual fluids administered revealed trends similar to those calculated using volumetric data. The amount of infused 0.9% saline was weakly correlated with the lowest blood pH and the highest serum chloride levels (ρ = -0.26 and 0.19, respectively). Conclusion Changes in the trends of fluid resuscitation practice were noted in a single surgical ICU over the 8-year study period. Crystalloid use increased owing to a rise in the utilization of balanced solutions with a downward trend in colloid use.

Background: Despite the increasing importance of rehabilitation for critically ill patients, there is little information regarding how rehabilitation therapy is utilized in clinical practice. Our objectives were to evaluate the implementation rate of rehabilitation therapy in the intensive care unit (ICU) survivors and to investigate the effects of rehabilitation therapy on outcomes. Methods: A retrospective nationwide cohort study with including > 18 years of ages admitted to ICU between January 2008 and May 2015 (n = 1,465,776). The analyzed outcomes were readmission to ICU readmission and emergency room (ER) visit. Results: During the study period, 249,918 (17.1%) patients received rehabilitation therapy. The percentage of patients receiving any rehabilitation therapy increased annually from 14% in 2008 to 20% in 2014, and the percentages for each type of therapy also increased over time. The most common type of rehabilitation was physical therapy (91.9%), followed by neuromuscular electrical stimulation (29.6%), occupational (28.6%), respiratory, (11.6%) and swallowing (10.3%) therapies. After adjusting for confounding variables, the risk of 30-day ICU readmission was lower in patients who received rehabilitation therapy than in those who did not (P < 0.001; hazard ratio [HR], 0.70; 95% confidence interval [CI], 0.65–0.75). And, the risk of 30-day ER visit was also lower in patients who received rehabilitation therapy (P < 0.001; HR, 0.83; 95% CI, 0.77–0.88). Conclusion In this nationwide cohort study in Korea, only 17% of all ICU patients received rehabilitation therapy. However, rehabilitation is associated with a significant reduction in the risk of 30-day ICU readmission and ER visit.

BACKGROUND: Beyond its current function as a rescue therapy in acute respiratory distress syndrome (ARDS), extracorporeal membrane oxygenation (ECMO) may be applied in ARDS patients with less severe hypoxemia to facilitate lung protective ventilation. The purpose of this study was to evaluate the efficacy of extended ECMO use in ARDS patients. METHODS: This study reviewed 223 adult patients who had been admitted to the intensive care units of 11 hospitals in Korea and subsequently treated using ECMO. Among them, the 62 who required ECMO for ARDS were analyzed. The patients were divided into two groups according to pre-ECMO arterial blood gas: an extended group (n=14) and a conventional group (n=48). RESULTS: Baseline characteristics were not different between the groups. The median arterial carbon dioxide tension/fraction of inspired oxygen (FiO2) ratio was higher (97 vs. 61, p<0.001) while the median FiO2 was lower (0.8 vs. 1.0, p<0.001) in the extended compared to the conventional group. The 60-day mortality was 21% in the extended group and 54% in the conventional group (p=0.03). Multivariate analysis indicated that the extended use of ECMO was independently associated with reduced 60-day mortality (odds ratio, 0.10; 95% confidence interval, 0.02–0.64; p=0.02). Lower median peak inspiratory pressure and median dynamic driving pressure were observed in the extended group 24 hours after ECMO support. CONCLUSION Extended indications of ECMO implementation coupled with protective ventilator settings may improve the clinical outcome of patients with ARDS.

BACKGROUND: Beyond its current function as a rescue therapy in acute respiratory distress syndrome (ARDS), extracorporeal membrane oxygenation (ECMO) may be applied in ARDS patients with less severe hypoxemia to facilitate lung protective ventilation. The purpose of this study was to evaluate the efficacy of extended ECMO use in ARDS patients. METHODS: This study reviewed 223 adult patients who had been admitted to the intensive care units of 11 hospitals in Korea and subsequently treated using ECMO. Among them, the 62 who required ECMO for ARDS were analyzed. The patients were divided into two groups according to pre-ECMO arterial blood gas: an extended group (n=14) and a conventional group (n=48). RESULTS: Baseline characteristics were not different between the groups. The median arterial carbon dioxide tension/fraction of inspired oxygen (FiO2) ratio was higher (97 vs. 61, p<0.001) while the median FiO2 was lower (0.8 vs. 1.0, p<0.001) in the extended compared to the conventional group. The 60-day mortality was 21% in the extended group and 54% in the conventional group (p=0.03). Multivariate analysis indicated that the extended use of ECMO was independently associated with reduced 60-day mortality (odds ratio, 0.10; 95% confidence interval, 0.02–0.64; p=0.02). Lower median peak inspiratory pressure and median dynamic driving pressure were observed in the extended group 24 hours after ECMO support. CONCLUSION: Extended indications of ECMO implementation coupled with protective ventilator settings may improve the clinical outcome of patients with ARDS.
Adult ; Anoxia ; Carbon Dioxide ; Extracorporeal Membrane Oxygenation ; Humans ; Intensive Care Units ; Korea ; Lung ; Mortality ; Multicenter Studies as Topic ; Multivariate Analysis ; Oxygen ; Respiration, Artificial ; Respiratory Distress Syndrome, Adult ; Retrospective Studies ; Ventilation ; Ventilators, Mechanical