Consecutive Intensive Care Unit Research Articles

Tele-Critical Care, Severity of Illness, and 30-Day Mortality Risk: A Retrospective, Cohort Analysis.

Introduction: Studies have shown that tele-critical care (TCC) improves outcomes in intensive care unit (ICU) settings with low baseline performance. Evidence also suggests that TCC outcomes may be modified by heterogenous baseline severity of illness. We examined the association of admission Acute Physiology and Chronic Health Evaluation IV (APACHE IV) score quartiles (APQ1-APQ4) and TCC exposure with 30-day mortality. Methods: Retrospective, cohort study of 151,780 consecutive ICU patients admitted to nine hospitals within Cleveland Clinic Health System from 2010 to 2019. Data were abstracted from an institutional ICU Datamart and APACHE IV registry. Analyses included summary statistics for demographics, unadjusted survival functions, and incidence rates across ascending APACHE quartiles (APQ1-APQ4). Multivariate Poisson regression modeled covariates associated with incidence rate ratio (IRR) for mortality, across quartiles including statistical interaction between TCC exposure and APACHE quartiles. Results: Unadjusted mortality risk ratios of TCC/no TCC were statistically different across APQ1 (0.83; 95% confidence interval [CI] 0.71-0.97), APQ3 (0.63; 95% CI 0.57-0.69), and APQ4 (0.77; 95% CI 0.74-0.82) (all p < 0.05) but not in APQ2 (0.98; 95% CI 0.88-1.10; p = 0.77). Multivariate Poisson modeling found reduced IRR with TCC (IRR 0.82; 95% CI 0.70-0.97). Relative to APQ1, risk was increased across quartiles, APQ2 (IRR 2.15; 95% CI 1.83-2.52), APQ3 (IRR 3.93; 95% CI 3.39-4.56), and APQ4 (IRR 9.30; 95% CI 8.10-10.67). Interaction with TCC significantly reduced risk in APQ3 (IRR 0.80; 95% CI 0.67-0.96). Conclusion: TCC exposure is associated with reduced 30-day mortality, affected by various clinical factors, to provide heterogenous impact. Mortality benefit appears to particularly accrue among patients with higher, but not the highest quartile for severity of illness, based on their APACHE IV scores.

End-of-life practices in Hong Kong intensive care units: results from the Ethicus-2 study.

The need for end-of-life care is common in intensive care units (ICUs). Although guidelines exist, little is known about actual end-of-life care practices in Hong Kong ICUs. The study aim was to provide a detailed description of these practices. This prospective, multicentre observational sub-analysis of the Ethicus-2 study explored end-of-life practices in eight participating Hong Kong ICUs. Consecutive adult ICU patients admitted during a 6-month period with life-sustaining treatment (LST) limitation or death were included. Follow-up continued until death or 2 months from the initial decision to limit LST. Of 4922 screened patients, 548 (11.1%) had LST limitation (withholding or withdrawal) or died (failed cardiopulmonary resuscitation/brain death). Life-sustaining treatment limitation occurred in 455 (83.0%) patients: 353 (77.6%) had decisions to withhold LST and 102 (22.4%) had decisions to withdraw LST. Of those who died without LST limitation, 80 (86.0%) had failed cardiopulmonary resuscitation and 13 (14.0%) were declared brain dead. Discussions of LST limitation were initiated by ICU physicians in most (86.2%) cases. Shared decision-making between ICU physicians and families was the predominant model; only 6.0% of patients retained decision-making capacity. Primary medical reasons for LST limitation were unresponsiveness to maximal therapy (49.2%) and multiorgan failure (17.1%). The most important consideration for decision-making was the patient's best interest (81.5%). Life-sustaining treatment limitations are common in Hong Kong ICUs; shared decision-making between physicians and families in the patient's best interest is the predominant model. Loss of decision-making capacity is common at the end of life. Patients should be encouraged to communicate end-of-life treatment preferences to family members/surrogates, or through advance directives.

Clinical significance of transjugular liver biopsy in acute liver failure – a real-world analysis

BackgroundHistopathological characterization obtained by transjugular liver biopsy (TJLB) may theoretically contribute to clarification of the exact aetiology of acute liver failure (ALF). It's unclear whether the histopathological information from TJLB, due to the small specimen size, significantly contributes to diagnosing ALF causes, guiding therapy decisions, or predicting overall prognosis. This retrospective study aimed to analyse safety and clinical significance of TJLB in patients with ALF.MethodsThis retrospective, monocentric study investigated safety and efficacy of TJLB in patients with ALF over a ten-year period at a tertiary care transplant-center. The predictive value of various clinical and laboratory characteristics as well as histopathological findings obtained by TJLB on 28-day liver-transplant-free survival were evaluated by calculating uni- and multivariate Cox-proportional hazard regression models. Additional univariate logistic regression analyses were performed to explore the influence of degree of intrahepatic necrosis on the secondary endpoints intensive-care-unit (ICU) admission, need for endotracheal intubation, renal replacement therapy and high-urgency listing for LTX.ResultsA total of 43 patients with ALF receiving TJLB were included into the study. In most cases (n = 39/43 cases) TJLB confirmed the initially already clinically presumed ALF aetiology and the therapeutic approach was unchanged by additional histological examination in the majority of patients (36/43 cases). However, in patients with a high suspicion for aetiologies potentially treatable by medical immunosuppression (e.g. AIH, GvHD), TJLB significantly influenced further treatment planning and/or adjustment. While the degree of intrahepatic necrosis showed significance in the univariate analysis (p = 0.04), it did not demonstrate a significant predictive effect on liver transplant-free survival in the multivariate analysis (p = 0.1). Only consecutive ICU admission was more likely with higher extent of intrahepatic necrosis (Odds ratio (OR) 1.04 (95% CI 1–1.08), p = 0.046).ConclusionsPerformance of TJLB in ALF led to a change in suspected diagnosis and to a significant change in therapeutic measures only in those patients with a presumed high risk for aetiologies potentially responsive to immunosuppressive therapy. Clinical assessment alone was accurate enough, with additional histopathological examination adding no significant value, to predict overall prognosis of patients with ALF.

Cardiac biomarkers for screening and prognostication of cardiac dysfunction in critically ill patients.

This study aimed to assess the use of high-sensitivity troponin T (hsTNT) and N-terminal pro-brain natriuretic peptide (NT-proBNP) in screening for cardiac dysfunction [left ventricular (LV) systolic or diastolic dysfunction or right ventricular (RV) dysfunction] in mixed intensive care unit (ICU) patients and establish whether these biomarkers are independently associated with an increased risk of death. We performed a secondary analysis of a single-centre prospective observational study in which consecutive ICU patients were examined with transthoracic echocardiography (TTE) and cardiac biomarkers. Patients with systolic or diastolic LV dysfunction, RV dysfunction or a combination of these were compared with patients with normal cardiac function. Sensitivity and specificity for different cut-off levels were calculated using receiver operating characteristic curves. Regression models were used to evaluate the associations between cardiac biomarkers, sepsis, renal failure and mortality. A total of 276 patients were included. Most of the patients had cardiac dysfunction on TTE (64%). Combined cardiac dysfunction was most prevalent (71 patients, 26%), followed by isolated diastolic LV dysfunction (40 patients, 15%). Levels of hsTNT and NT-proBNP were higher in all types of cardiac dysfunction versus patients with normal cardiac function. The area under the curve (AUC) for hsTNT to detect any cardiac dysfunction was 0.75. An optimal cut-off at 30.5ng/L rendered a positive predictive value (PPV) of 80% and a negative predictive value (NPV) of 58%. The AUC for NT-proBNP to detect any cardiac dysfunction was 0.788. Using an optimal cut-off at 1145ng/L rendered a PPV of 86% and an NPV of 58%. Using a clinically relevant 90% sensitivity for detecting cardiac dysfunction put the cut-offs at 14.1ng/L for hsTNT and 247ng/L for NT-proBNP, resulting in a specificity of 48% and 46%, respectively. Levels of NT-proBNP were associated with sepsis and renal failure (P<0.001), while levels of hsTNT were associated with renal failure only (P<0.001) after adjustment for cardiac dysfunction. Levels of biomarkers were associated with an increased risk of 90day mortality after adjustments for age, Simplified Acute Physiology Score 3, cardiac dysfunction and factors independently associated with biomarker increase (sepsis and renal failure) (P=0.048 for hsTNT and P<0.006 for NT-proBNP). Cardiac biomarkers, hsTNT and NT-proBNP, are strongly correlated to cardiac dysfunction in ICU patients and have a robust association with increased mortality. However, the relatively low NPV and the low specificity at relevant sensitivity levels of the biomarkers make them unsuitable for use in screening for cardiac dysfunction.

Prospective multicenter study to describe the prevalence, outcomes, and management of phosphate disorders in intensive care patients: Study protocol for part B of the international GUTPHOS study

BackgroundAberrations in blood phosphate (Pi) levels, whether presenting as hypo- or hyperphosphatemia, appear to be associated with clinical complications and adverse outcomes in patients admitted to an intensive care unit (ICU). However, the prevalence of Pi disorders and the association with subsequent factors and organ failures leading to death in ICU patients are poorly described. Despite endeavors to understand the etiology and treatment of low Pi levels from systematic reviews and meta-analyses, the literature lacks comprehensive guidance for managing hypophosphatemia. Hyperphosphatemia, on the other hand, appears to be associated with higher mortality among critically ill patients, yet its prevalence among ICU patients, particularly following phosphate repletion, remains unknown. The present study aims to investigate the prevalence of Pi abnormalities upon ICU admission and their incidence during the first week of ICU stay, the factors associated with Pi alterations, and the effect of phosphate repletion on the normalization of Pi levels, and its associations with clinical outcomes. MethodsThis multicentre, prospective, non-interventional cohort study will include at least 1000 consecutive adult ICU patients (≥18 years) as part B of the GUTPHOS study. Sites are eligible if an anticipated minimal inclusion of 50 eligible patients during eight weeks from January 2024 until June 2024 and daily phosphate measurements during the first seven days of ICU stay are expected. All consecutive adult patients admitted to a participating ICU during the recruitment period, lasting up to eight weeks, or up to 120 patients if enrollment reaches that limit earlier, will be included. Study parameters include study site characteristics, patient demographics, daily assessment of Pi levels, Pi-related treatment, feeding details, renal replacement therapy details, the incidence of refeeding-associated hypophosphatemia and administered medication (during the first seven calendar days of ICU stay). There will be a follow-up period of a maximum of 90 days to document 28- and 90-day all-cause mortality as the primary outcome. Multiple logistic regression will be used to assess independent associations with mortality in addition to Receiver Operating Characteristics curves to identify cut-off Pi values associated with mortality and overcorrection. Linear mixed models will be conducted to assess Pi treatment effects. Subgroup analyses will be performed based on Pi abnormalities observed during ICU admission, categorized as normo-, hypo-, hyper-, or mixed, along with its severity (mild, moderate, or severe). DiscussionThe GUTPHOS study will be the first multicentre, prospective observational cohort study to investigate the prevalence, management practices, and consequent outcomes associated with Pi abnormalities during the first week of ICU admission. Its results may bridge the current evidence gap in repletion protocols while establishing the groundwork for a subsequent randomized controlled trial. Clinical trial registryNCT05909722.

Social Determinants of Health and Limitation of Life-Sustaining Therapy in Neurocritical Care: A CHoRUS Pilot Project.

Social determinants of health (SDOH) have been linked to neurocritical care outcomes. We sought to examine the extent to which SDOH explain differences in decisions regarding life-sustaining therapy, a key outcome determinant. We specifically investigated the association of a patient's home geography, individual-level SDOH, and neighborhood-level SDOH with subsequent early limitation of life-sustaining therapy (eLLST) and early withdrawal of life-sustaining therapy (eWLST), adjusting for admission severity. We developed unique methods within the Bridge to Artificial Intelligence for Clinical Care (Bridge2AI for Clinical Care)Collaborative Hospital Repository Uniting Standards for Equitable Artificial Intelligence (CHoRUS)program to extract individual-level SDOH from electronic health records and neighborhood-level SDOH from privacy-preserving geomapping. We piloted these methods to a 7years retrospective cohort of consecutive neuroscience intensive care unit admissions (2016-2022) at two large academic medical centers within an eastern Massachusetts health care system, examining associations between home census tract and subsequent occurrence of eLLST and eWLST. We matched contextual neighborhood-level SDOH information to each census tract using public data sets, quantifying Social Vulnerability Index overall scores and subscores. We examined the association of individual-level SDOH and neighborhood-level SDOH with subsequent eLLST and eWLST through geographic, logistic, and machine learning models, adjusting for admission severity using admission Glasgow Coma Scale scores and disorders of consciousness grades. Among 20,660 neuroscience intensive care unit admissions (18,780 unique patients), eLLST and eWLST varied geographically and were independently associated with individual-level SDOH and neighborhood-level SDOH across diagnoses. Individual-level SDOH factors (age, marital status, and race) were strongly associated with eLLST, predicting eLLST more strongly than admission severity. Individual-level SDOH were more strongly predictive of eLLST than neighborhood-level SDOH. Across diagnoses, eLLST varied by home geography and was predicted by individual-level SDOH and neighborhood-level SDOH more so than by admission severity. Structured shared decision-making tools may therefore represent tools for health equity. Additionally, these findings provide a major warning: prognostic and artificial intelligence models seeking to predict outcomes such as mortality or emergence from disorders of consciousness may be encoded with self-fulfilling biases of geography and demographics.

Assessing pain in paralyzed critically ill patients receiving neuromuscular blocking agents: A monocenter prospective cohort

IntroductionGiven the absence of established recommendations for pain assessment in pharmacologically paralyzed Intensive-Care-Units (ICU) patients under Neuro-Muscular-Blocking Agents (NMBA), this study assessed the validity of various parameters for evaluating pain in this specific population. Patients and methodsFour electrophysiological parameters (instant-Analgesia-Nociception-Index (ANI), Bispectral index (BIS), Heart Rate (HR) and Mean Arterial Blood Pressure (ABP)) and one clinical parameter (Behavioural-Pain-Scale (BPS)) were recorded during tracheal-suctioning in all consecutive ICU patients who required a continuous infusion of cisatracurium, before and just after paralysis recovery measured by Train-of-Four ratio. The validity of the five pain-related parameters was assessed by comparing the values recorded during different situations (before/during/after the nociceptive procedure) (discriminant-validity, primary outcome), and the effect of paralysis was assessed by comparing values obtained during and after paralysis (reliability, secondary outcome). ResultsTwenty patients were analyzed. ANI, BIS, and HR significantly changed during the nociceptive procedure in both paralysis and recovery, while BPS changed only post-recovery. ANI and HR were unaffected by paralysis, unlike BIS and BPS (mixed-effect model). ANI exhibited the highest discriminant-validity, with values (min 0/max 100) decreasing from 71 [48–89] at rest to 41 [25–72] during tracheal suctioning in paralyzed patients, and from 71 [53–85] at rest to 40 [31–52] in non-paralyzed patients. ConclusionsANI proves the most discriminant parameter for pain detection in both paralyzed and non-paralyzed sedated ICU patients. Its significant and clinically relevant decrease during tracheal suctioning remains unaltered by NMBA use. Pending further studies on analgesia protocols based on ANI, it could be used to assess pain during nociceptive procedures in ICU patients receiving NMBA.

Dosing Regimen for Cefotaxime Should Be Adapted to the Stage of Renal Dysfunction in Critically Ill Adult Patients-A Retrospective Study.

Cefotaxime administration is recommended in doses of 3-12 g/day in adults with a Glomerular Filtration Rate (GFR) > 5 mL/min. This study aimed to assess the impact of renal function and obesity on cefotaxime concentrations in intensive care unit (ICU) patients. A retrospective cohort study was conducted on consecutive ICU patients receiving continuous cefotaxime infusion between 2020 and 2022 [IRBN992021/CHUSTE]. Doses were not constant; consequently, a concentration-to-dose ratio (C/D) was considered. Statistical analysis was performed to assess the relationship between cefotaxime concentrations, renal function, and obesity. A total of 70 patients, median age 61 years, were included, with no significant difference in cefotaxime concentrations between obese and non-obese patients. However, concentrations varied significantly by GFR, with underdosing prevalent in patients with normal to increased renal function and overdosing in those with severely impaired renal function. Adjustment of cefotaxime dosing according to GFR was associated with improved target attainment. Cefotaxime dosing in critically ill patients should consider renal function, with higher initial doses required in patients with normal to increased GFR and lower doses in those with severely impaired renal function. Therapeutic drug monitoring may aid in optimising dosing regimens. Prospective studies are warranted to validate these findings and inform clinical practice.

Epidemiology and Outcomes of Critical Illness and Novel Predictors of Mortality in an Ethiopian Medical Intensive Care Unit.

Low- and middle-income countries (LMICs) bear most of the global burden of critical illness. Managing this burden requires improved understanding of epidemiology and outcomes in LMIC intensive care units (ICUs), including LMIC-specific mortality prediction scores. This study was a retrospective observational study at Tikur Anbessa Specialized Hospital in Addis Ababa, Ethiopia, examining all consecutive medical ICU admissions from June 2014 to April 2015. The primary outcome was ICU mortality; secondary outcomes were prolonged ICU stay and prolonged mechanical ventilation. ICU mortality prediction models were created using multivariable logistic regression and compared with the Mortality Probability Model-II (MPM-II). Associations with secondary outcomes were examined with multivariable logistic regression. There were 198 admissions during the study period; mortality was 35%. Age, shock on admission, mechanical ventilation, human immunodeficiency virus, and Glasgow Coma Scale ≤8 were associated with ICU mortality. The receiver operating characteristic curve for this 5-factor model had an AUC of 0.8205 versus 0.7468 for MPM-II, favoring the simplified new model. Mechanical ventilation and lack of shock were associated with prolonged ICU stays. Mortality in an LMIC medical ICU was high. This study examines an LMIC medical ICU population, showing a simplified prediction model may predict mortality as well as complex models.

Abstract 13943: Comparison of the Frequency and Time to True Ventricular Tachycardia Among Patients in the Cardiac, Medical/Surgical and Neurologic Intensive Care Units

Introduction: Continuous ECG monitoring is the non-invasive gold standard used to identify ventricular tachycardia (VT), yet 87% of current bedside monitor VTs could be false. Therefore, knowing the burden of true VT is essential to guide treatment to those at greatest risk of poor outcomes. Hypothesis: When stratifying by intensive care unit (ICU) (cardiac, medical/surgical, neurological), there will be differences in: 1) demographic and clinical characteristics of patients with true VT; and 2) frequency and time to first true VT. Methods: Retrospective cohort study of adults admitted to an academic medical center ICU, 9/13-4/15. Potential VTs among 5,679 consecutive ICU patients were manually annotated by clinicians as true/false. Patient-level data were extracted from the health record. We used Kaplan-Meier and Cox proportional hazards methods to calculate time to first true VT and 60-day VT rate by ICU type. Results: There were differences in all characteristics examined (p&lt;0.0001)(Table) and frequency of true VT, stratified by ICU type (log-rank p&lt;0.0001)(Figure). After adjustment for age, gender, and risk factors (Table), CICU patients had a 42% increased rate (HR 1.42, 95% CI 1.16-1.74) and NICU patients had a 28% decreased rate of VT (HR 0.72, 95% CI 0.59-0.89) compared to M/SICU patients. Conclusion: Patients in the CICU experienced true VT much earlier (1.5 days) after admission than patients in the M/SICU and NICU. In addition, CICU patients had a higher rate of true VT and associated VT risk factors as compared to M/SICU and NICU patients. As expected, known cardiac risk factors are associated with true VT, but may be hidden among false alarms during bedside monitoring.

Outcomes of patients admitted to the intensive care unit with community-acquired pneumonia in a tertiary care center in Riyadh, Saudi Arabia.

Community-acquired pneumonia (CAP) is a leading cause of intensive care unit (ICU) morbidity and mortality. Despite extensive international epidemiological and clinical studies to improve those patients' outcomes, local statistics in Saudi Arabia are limited. The objective of this study is to describe the clinical characteristics and outcomes of patients admitted to the ICU with the diagnosis of CAP reflecting the experience of a tertiary center over an 18-year period. A retrospective cohort study included all consecutive adult ICU patients diagnosed with CAP between 1999 and 2017. Baseline demographics, patients' risk factors, and initial admission laboratory investigations were compared between survivors and nonsurvivors. A multivariate regression model was used to predict mortality. During the study period, there were 3438 patients admitted to the ICU with CAP (median age 67 [Quartile 1, 3 (Q1, Q3) 51, 76] years) and 54.4% were males, of whom 1007 (29.2%) died. The survivors compared with nonsurvivors were younger (65 vs. 70 years), less likely to have chronic liver disease (2.4% vs. 10.5%), chronic renal failure (8.1% vs. 14.4%), and be immunocompromised (10.2% vs. 18.2%), and less frequently required mechanical ventilation or vasopressors (46.2% vs. 80.5% and 29.6% vs. 55.9%, respectively). Acute Physiology and Chronic Health Evaluation (APACHE) II score was significantly higher among nonsurvivors (median score 26 vs. 20) with a longer duration of mechanical ventilation and ICU stay. Using a multivariate regression model, age, APACHE II score, bilirubin level, vasopressors, and mechanical ventilation were significantly associated with increased mortality, while diabetes was associated with lower mortality. Around one-third of patients admitted to the ICU with CAP died. Mortality was significantly associated with age, APACHE II score, vasopressor use, and mechanical ventilation. A comprehensive national registry is needed to enhance epidemiological data and to guide initiatives for improving CAP patients' outcomes.

Ventricular tachycardia and in-hospital mortality in the intensive care unit

Continuous electrocardiographic (ECG) monitoring is used to identify ventricular tachycardia (VT), but false alarms occur frequently. The purpose of this study was to assess the rate of 30-day in-hospital mortality associated with VT alerts generated from bedside ECG monitors to those from a new algorithm among intensive care unit (ICU) patients. We conducted a retrospective cohort study in consecutive adult ICU patients at an urban academic medical center and compared current bedside monitor VT alerts, VT alerts from a new-unannotated algorithm, and true-annotated VT. We used survival analysis to explore the association between VT alerts and mortality. We included 5679 ICU admissions (mean age 58 ± 17 years; 48% women), 503 (8.9%) experienced 30-day in-hospital mortality. A total of 30.1% had at least 1 current bedside monitor VT alert, 14.3% had a new-unannotated algorithm VT alert, and 11.6% had true-annotated VT. Bedside monitor VT alert was not associated with increased rate of 30-day mortality (adjusted hazard ratio [aHR] 1.06; 95% confidence interval [CI] 0.88-1.27), but there was an association for VT alerts from our new-unannotated algorithm (aHR 1.38; 95% CI 1.12-1.69) and true-annotated VT(aHR 1.39; 95% CI 1.12-1.73). Unannotated and annotated-true VT were associated with increased rate of 30-day in-hospital mortality, whereas current bedside monitor VT was not. Our new algorithm may accurately identify high-risk VT; however, prospective validation is needed.

Extracorporeal Blood Purification with CytoSorb in 359 Critically Ill Patients

Introduction: Critically ill patients with inflammatory dysregulation and organ disfunction may benefit from blood purification, although the use of this technique has not been described in large case series. We evaluated clinical outcomes and survival in high-risk intensive care unit (ICU) patients who underwent extracorporeal blood purification. Methods: 359 consecutive ICU patients treated with CytoSorb were included. Results: Main admission diagnoses were 120 (34%) refractory cardiac arrest under mechanical chest compression; 101 (28%) profound cardiogenic shock; 81 (23%) post-cardiotomy cardiogenic shock; and 37 (10%) respiratory failure. Fifteen patients (4%) were positive for SARS-CoV-2 infection. We observed 49% 30-day mortality, 57% ICU mortality, and 62% hospital mortality, all lower than the 71% mortality predicted by SAPS II and 68% predicted by SOFA score. Parameters of shock and organ failure, above all vasoactive inotropic score, reduced during CytoSorb treatment. Multivariable analysis identified SAPS II, lactate dehydrogenase, ICU stay duration, vasoactive inotropic score, lactates, intra-aortic counterpulsation on top of VA-ECMO, and total bilirubin as predictors of mortality. No CytoSorb-related complications occurred. Conclusion: CytoSorb treatment was effective in reducing laboratory parameters of shock and vasoactive inotropic score with possible survival implications in a large population of critically ill patients.

Thromboembolic Presentations among Patients Hospitalized to the Intensive Care Unit for Coronavirus Disease 2019 (COVID-19) - A Northern Taiwan Single Center Experience.

Coronavirus Disease 2019 (COVID-19) has been associated with a high thromboembolic risk among patients in intensive care units. Asian populations may share a similar thromboembolic risk, but with a higher prevalence of arterial thromboembolism than venous thromboembolism. To clarify this risk in Taiwan, this single-center retrospective study collected 27 consecutive intensive care unit patients with COVID-19 confirmed by polymerase chain reaction, with a median age of 67.6 years (male 81.5%). Twenty-three patients received prophylactic anticoagulation (85.2%), and there were four bleeding events (14.8%). Nine patients had thromboembolism (33.3%), including three with deep vein thrombosis, two with peripheral artery thromboembolism, and four with ischemic stroke. There were no significant clinical differences between the patients with or without thromboembolism. Initial serum ferritin [adjusted odds ratio (OR): 13.19, 95% confidence interval (CI): 1.01-172.07] and peak serum procalcitonin (adjusted OR: 18.93, 95% CI: 1.08-330.91) were associated with a higher risk of thromboembolism. Furthermore, prophylactic anticoagulation (adjusted OR: 0.01, 95% CI: < 0.001-0.55) was associated with a lower risk of thromboembolism. All cases of deep vein thrombosis and one peripheral artery thromboembolism occurred at intravascular catheter locations. No association between thromboembolism and survival was found (age-adjusted hazard ratio: 0.55, 95% CI: 0.10-2.95). In conclusion, the prevalence of COVID-19 thromboembolism among Taiwanese patients in intensive care units was high, even with prophylactic anticoagulation. Serum ferritin and procalcitonin may identify high-risk populations. Prophylactic anticoagulation may reduce the risk of thromboembolism with a manageable bleeding risk. Larger prospective studies are needed to clarify the risk of COVID-19 thromboembolism and its risk factors in the post-Omicron era.

Association of perinephric fat stranding with CT severity scores and mortality in COVID-19 patients

Aims: To investigate the relationship between the presence of perinephric fat stranding (PFS) with three consecutive chest CT severity scores (CT-SS), mortality and intensive care unit admission in COVID-19 patients. Methods: This single-center retrospective study, included 457 (?18 years) COVID-19 patients with positive RT-PCR tests. A radiologist calculated three consecutive CT-SS for COVID-19 pneumonia using a visual scoring system ranging from 0 to 25 points. Grades of PFS on CTs were defined as none, mild, moderate or severe. Firstly, patients were divided into two groups, with and without PFS. The Association of PFS with demographic and laboratory data, CT-SS, and mortality rates were analyzed. We divided patients into four subgroups based on PFS grading and investigated temporal changes of mean CT-SS of three consecutive CTs in each PFS subgroup using a paired-sample test and Wilcoxon signed-rank test. Results: Patients with PFS were associated with older age (p&lt;0.001) and had higher CT-SS (p=0.03). We found a particularly strong association between PFS and mortality (p&lt;0.001) and intensive care unit admission (p=0.001). Statistical associations were found between PFS and elevated serum BUN (p=0.004), creatinine (p=0.007), CRP (p=0.02), and ferritin (p=0.005). In multivariate logistic regression analysis, older age was associated with 1.067-fold (p&lt;=0.001), PFS 1.964-fold (p=0.007), elevated serum creatinine 3.630-fold (p=0.005) higher risks of mortality. In PFS subgroups other than severe, there were significant increases between the first and second CT-SS (p&lt;0.001, p&lt;0.001, p=0.003). Conclusion: Perinephric fat stranding is an important CT finding that can alert clinicians to the poor prognosis of COVID-19 patients in early periods.

Continuous and deep sedation until death after a decision to withdraw life-sustaining therapies in intensive care units: A national survey

Background: Continuous and deep sedation until death is a much highly debated end-of-life practice. France is unique in having a regulatory framework for it. However, there are no data on its practice in intensive care units (ICUs). Aim: The aim is to describe continuous and deep sedation in relation to the framework in the specific context of withdrawal of life-sustaining therapies in ICUs, that is, its decision-making process and its practice compared to other end-of-life practices in this setting. Design and setting: French multicenter observational study. Consecutive ICU patients who died after a decision to withdraw life-sustaining therapies. Results: A total of 343 patients in 57 ICUs, 208 (60%) with continuous and deep sedation. A formalized procedure for continuous and deep sedation was available in 32% of the ICUs. Continuous and deep sedation was not the result of a collegial decision-making process in 17% of cases, and did not involve consultation with an external physician in 29% of cases. The most commonly used sedative medicines were midazolam (10 [5–18] mg h−1) and propofol (200 [120–250] mg h −1). The Richmond Agitation Sedation Scale (RASS) was −5 in 60% of cases. Analgesia was associated with sedation in 94% of cases. Compared with other end-of-life sedative practices (n = 98), medicines doses were higher with no difference in the depth of sedation. Conclusions: This study shows a poor compliance with the framework for continuous and deep sedation. It highlights the need to formalize it to improve the decision-making process and the match between the intent, the practice and the actual effect.

End-of-life practices in 11German intensive care units : Results from the ETHICUS-2 study.

End-of-life care is common in German intensive care units (ICUs) but little is known about daily practice. To study the practice of end-of-life care. Prospectively planned, secondary analysis comprising the German subset of the worldwide Ethicus‑2 Study (2015-2016) including consecutive ICU patients with limitation of life-sustaining therapy or who died. Among 1092 (13.7%) of 7966 patients from 11multidisciplinary ICUs, 967 (88.6%) had treatment limitations, 92 (8.4%) died with failed CPR, and 33 (3%) with brain death. Among patients with treatment limitations, 22.3% (216/967) patients were discharged alive from the ICU. More patients had treatments withdrawn than withheld (556 [57.5%] vs. 411 [42.5%], p < 0.001). Patients with treatment limitations were older (median 73years [interquartile range (IQR) 61-80] vs. 68years [IQR 54-77]) and more had mental decision-making capacity (12.9vs. 0.8%), advance directives (28.6vs. 11.2%), and information about treatment wishes (82.7 vs 33.3%, all p < 0.001). Physicians reported discussing treatment limitations with patients with mental decision-making capacity and families (91.3and 82.6%, respectively). Patient wishes were unknown in 41.3% of patients. The major reason for decision-making was unresponsiveness to maximal therapy (34.6%). Treatment limitations are common, based on information about patients' wishes and discussion between stakeholders, patients and families. However, our findings suggest that treatment preferences of nearly half the patients remain unknown which affects guidance for treatment decisions.

Critical Care Cardiology Trials Network (CCCTN): a cohort profile.

The aims of the Critical Care Cardiology Trials Network (CCCTN) are to develop a registry to investigate the epidemiology of cardiac critical illness and to establish a multicentre research network to conduct randomised clinical trials (RCTs) in patients with cardiac critical illness. The CCCTN was founded in 2017 with 16 centres and has grown to a research network of over 40 academic and clinical centres in the United States and Canada. Each centre enters data for consecutive cardiac intensive care unit (CICU) admissions for at least 2 months of each calendar year. More than 20000 unique CICU admissions are now included in the CCCTN Registry. To date, scientific observations from the CCCTN Registry include description of variations in care, the epidemiology and outcomes of all CICU patients, as well as subsets of patients with specific disease states, such as shock, heart failure, renal dysfunction, and respiratory failure. The CCCTN has also characterised utilization patterns, including use of mechanical circulatory support in response to changes in the heart transplantation allocation system, and the use and impact of multidisciplinary shock teams. Over years of multicentre collaboration, the CCCTN has established a robust research network to facilitate multicentre registry-based randomised trials in patients with cardiac critical illness. The CCCTN is a large, prospective registry dedicated to describing processes-of-care and expanding clinical knowledge in cardiac critical illness. The CCCTN will serve as an investigational platform from which to conduct randomised controlled trials in this important patient population.

P1363: OUTCOME OF ADULT HEMATOPOIETIC TRANSPLANT RECIPIENTS AFTER ADMISSION TO INTENSIVE CARE UNIT (ICU): SINGLE CENTRE EXPERIENCE.

Background: Intensive care unit (ICU) admission during hematopoietic stem cell transplant (HSCT) is associated with poor prognosis1,2. Published series report a range of ICU admission rates from 24-40% of transplant patients, most frequent reasons involving septic shock, respiratory failure and veno-occlusive disease3. In addition, patients undergoing HSCT are at a high risk of severe morbidity and mortality associated with COVID-194. Aims: The aim of this study was to analyze outcome of HSCT patients requiring ICU admission in our center. Methods: We retrospectively analysed outcome of 752 patients who underwent HSCT in our centre from January/2008 to June/2021. Data were collected from patients’ clinical histories. Results: 103 (14%) patients required ICU admission (baseline and HSCT characteristics on table). Median time to ICU admission was 42 days (-2-1765). Seven of these patients were admitted to ICU on two occasions giving a total of 110 consecutive ICU admissions available for analysis. Main reason for ICU admission was respiratory distress (74; 67%), mainly due to pneumonia (53%) including a 3% caused by COVID19, pulmonary edema (26%) and pulmonary haemorrhage (8%). Septic shock was second most common cause for ICU admission (26; 24%) due to gram-negative bacilli (47%), fungal (15%) gram-positive bacteria (13%), virus (10%) and others/idiopathic (16%). Other less frequent causes were veno-occlusive disease (11; 10%), hepatic failure/encephalopathy (8; 7%), haemorrhagic complications (6; 5%), cardiorespiratory arrest (2%), GVHD (2%), cardiogenic shock (2%). Of the 110 ICU admissions, 37 (34%) required hemofiltration, of which 30 (81%) died; and 77 (70%) required orotracheal intubation, of which 59 (77%) died. During the 110 ICU admissions, 67 patients (61%) died in the ICU; of these, 40 (37%) received unrelated donor HSCT, 36 (33%) sibling donor, 16 (15%) haploidentical and 17 (16%) autologous. Median ICU length of stay of these patients was 13 days (range 1-76). The cause of death was the same reason for ICU admission. Eighteen (16%) patients were discharged from ICU and died prior to hospital discharge and 24 (22%) survived to hospital discharge and were classified as post-discharge survivors. Of these 24 cases, 19 (79%) remain alive while the others (5; 21%) succumbed to underlying disease or complications post-HSCT. Off note, both patients with COVID19 pneumonia (haploidentical and autologous HSCT respectively) were discharged from ICU and remain alive to date, without major complications. Image:Summary/Conclusion: In our study 14% of transplant recipients required ICU admission, slightly lower than previous reports. Most common cause of admission was respiratory failure, consistent with reported. Mortality rate during ICU admission was 61%; higher death rate observed in allogeneic transplantation and those requiring aggressive ICU treatments such as mechanical ventilation or hemofiltration. Although patients with COVID19 pneumoniae who require ICU admission are usually associated with adverse outcome, in our series they responded successfully to intensive treatment. ICU admission following HSCT is associated with poor prognosis, but should not be considered futile.

Explainable Machine Learning-Based Risk Prediction Model for In-Hospital Mortality after Continuous Renal Replacement Therapy Initiation.

In this study, we established an explainable and personalized risk prediction model for in-hospital mortality after continuous renal replacement therapy (CRRT) initiation. This retrospective cohort study was conducted at Changhua Christian Hospital (CCH). A total of 2932 consecutive intensive care unit patients receiving CRRT between 1 January 2010, and 30 April 2021, were identified from the CCH Clinical Research Database and were included in this study. The recursive feature elimination method with 10-fold cross-validation was used and repeated five times to select the optimal subset of features for the development of machine learning (ML) models to predict in-hospital mortality after CRRT initiation. An explainable approach based on ML and the SHapley Additive exPlanation (SHAP) and a local explanation method were used to evaluate the risk of in-hospital mortality and help clinicians understand the results of ML models. The extreme gradient boosting and gradient boosting machine models exhibited a higher discrimination ability (area under curve [AUC] = 0.806, 95% CI = 0.770–0.843 and AUC = 0.823, 95% CI = 0.788–0.858, respectively). The SHAP model revealed that the Acute Physiology and Chronic Health Evaluation II score, albumin level, and the timing of CRRT initiation were the most crucial features, followed by age, potassium and creatinine levels, SPO2, mean arterial pressure, international normalized ratio, and vasopressor support use. ML models combined with SHAP and local interpretation can provide the visual interpretation of individual risk predictions, which can help clinicians understand the effect of critical features and make informed decisions for preventing in-hospital deaths.