Intensive Care Unit Setting Research Articles

Noise in a UK intensive care unit: a retrospective cohort study

Excessive noise in hospital Intensive care units (ICU) has been associated with adverse health effects and human factor impacts acknowledged to negatively affect both patients and healthcare providers. Aims: Evaluation of the acoustic environment of the Royal Liverpool University Hospital (RLUH) ICU versus WHO and UK guidelines. Methods: Noise measurements were taken from a six-bedded bay within the RLUH ICU between 15th June and 1st July 2022. This audit focussed on noise data > 87dBA, in accordance with the national exposure limit established by UK Noise Regulations. A retrospective review of relevant patients' records was conducted to identify potential noisy events during the audit period. Results: The highest and lowest LAeq value was 77.9 and 63.5dBA, respectively. All LAeq and LAmax noise measurements recorded exceeded recommended guidelines. Noise fell below recommended daytime guidelines less than 1% of the time. The top potential causes of noise were patient repositioning/personal care, medication administration and suctioning. Conclusion: Noise levels in the RLUH ICU considerably exceed national and international guidelines. These findings highlight the need to address noise pollution in the ICU setting. Hospital staff and policy makers should consider implementing strategies and interventions for noise reduction in ICUs.

Identifying risk factors for acute respiratory distress syndrome in critically ill patients: a retrospective study.

Acute respiratory distress syndrome (ARDS) is a life-threatening condition that can develop in critically ill patients. Early identification of risk factors associated with ARDS development is essential for timely intervention and improved patient outcomes. This study aimed to investigate the potential predictors of ARDS in critically ill patients admitted to the intensive care unit (ICU). We conducted a retrospective study involving 502 critically ill patients admitted to the ICUs of three hospitals. Demographic and clinical data, including laboratory test results, were collected during their ICU stay. Multivariable logistic regression analysis was performed to identify independent risk factors associated with the development of ARDS. Among the 502 critically ill patients, 104 (20.7%) patients developed ARDS during their ICU stay, with a median time to development of 5.2 days. Multivariable logistic regression analysis revealed that age (odds ratio [OR], 1.07; 95% confidence interval [CI], 1.01-1.13; P = 0.002), C-reactive protein (CRP) levels (OR, 1.11; 95% CI, 1.05-1.17; P = 0.013), T lymphocyte count (OR, 0.82; 95% CI, 0.69-0.93; P = 0.011), and interleukin-6 (IL-6) levels (OR, 1.17; 95% CI, 1.08-1.23; P = 0.003) were independently associated with the development of ARDS in critically ill patients. Our study identified age, CRP, T lymphocyte count, and IL-6 as independent predictors of ARDS in critically ill patients admitted to the ICU. These findings highlight the importance of monitoring these parameters in critically ill patients to identify those at high risk of developing ARDS. Early recognition and intervention based on these risk factors may improve patient outcomes in the ICU setting. Further prospective studies are warranted to validate these results and develop a reliable predictive model for ARDS in critically ill patients.

Non-invasive positive pressure ventilation for acute asthma in children.

Asthma is one of the most common reasons for hospital admission among children, with significant economic burden and impact on quality of life. Non-invasive positive pressure ventilation (NPPV) is increasingly used in the care of children with acute asthma, although the evidence supporting it is weak, and clinical guidelines do not offer any recommendations on its routine use. However, NPPV might be an effective way to improve outcomes for some children with asthma. A previous review did not demonstrate a clear benefit, but was limited by few studies with small sample sizes. This is an update of the previous review. To assess the benefits and harms of NPPV as an add-on therapy to usual care (e.g. bronchodilators and corticosteroids) in children (< 18 years) with acute asthma. We searched the Cochrane Airways Group Specialised Register, CENTRAL, MEDLINE, and Embase. We also conducted a search of ClinicalTrials.gov and the WHO ICTRP. We searched all databases from their inception to March 2023, with no restrictions on language of publication. We included randomised clinical trials (RCTs) assessing NPPV as add-on therapy to usual care versus usual care for children hospitalised for acute asthma exacerbations. We used standard Cochrane methods. We included three RCTs randomising 60 children with acute asthma to NPPV and 60 children to control. All included trials assessed the effects of bilevel positive airway pressure (BiPAP) for acute asthma in a paediatric intensive care unit (PICU) setting. None of the trials used continuous positive airway pressure (CPAP). The controls received standard care. The median age of children ranged from three to six years, and asthma severity ranged from moderate to severe. Our primary outcome measures were all-cause mortality, serious adverse events, and asthma symptom score. Secondary outcomes were non-serious adverse events, health-related quality of life, arterial blood gases and pH, pneumonia, cost, and PICU length of stay. None of the trials reported any deaths or serious adverse events (except one trial that reported intubation rate). Two trials reported asthma symptom score, each demonstrating reductions in asthma symptoms in the BiPAP group. In one trial, the asthma symptom score was (mean difference (MD) -2.50, 95% confidence interval (CI) -4.70 to -0.30, P = 0.03; 19 children) lower in the BiPAP group. In the other trial, a cross-over trial, BiPAP was associated with a lower mean asthma symptom score (MD -3.7; 16 children; very low certainty evidence) before cross-over, but investigators did not report a standard deviation, and it could not be estimated from the first phase of the trial before cross-over. The reduction in both trials was above our predefined minimal important difference. Overall, NPPV with standard care may reduce asthma symptom score compared to standard care alone, but the evidence is very uncertain. The only reported serious adverse event was intubation rate in one trial. The trial had an intubation rate of 40% and showed that BiPAP may result in a large reduction in intubation rate (risk ratio 0.47, 95% CI 0.23 to 0.95; 78 children), but the evidence is very uncertain. Post hoc analysis showed that BiPAP may result in a slight decrease in length of PICU stay (MD -0.87 day, 95% CI -1.52 to -0.22; 100 children), but the evidence is very uncertain. Meta-analysis or Trial Sequential Analysis was not possible because of insufficient reporting and different scoring systems. All three trials had high risk of bias with serious imprecision of results, leading to very low certainty of evidence. The currently available evidence for NNPV is uncertain. NPPV may lead to an improvement in asthma symptom score, decreased intubation rate, and slightly shorter PICU stay; however, the evidence is of very low certainty. Larger RCTs with low risk of bias are warranted.

Validity of a new weaning index of discontinuation from mechanical ventilation in intensive care unit

Background Failure to wean from mechanical ventilation (MV) is a frequent problem in the intensive care unit (ICU) setting. Herein, we investigate a new WI measured at the initial spontaneous breathing trial (SBT) in order to predict successful liberation from MV in mechanically ventilated ICU cases. Patients and methods This prospective cross-sectional study consisted of two phases; Phase I was designed to create a new WI from the criteria of 60 mechanically ventilated patients, while Phase II aimed to assess the accuracy of that index in another 40 patients. Results The incidence of failed weaning was 33.3% in phase I, and 35% in phase II. Multivariate regression analysis was done for phase I variables, and the following equation was created; (((Cough strength×0.743)–1.927)×0.05+((PaCO2 (mmHg)×-0.126)+6.786)×0.08+((mean airway pressure (CmH2O)×-1.304)+14.119)×0.193+((PaO2/FIO2 × 0.268)-58.555)×0.328+((rapid shallow breathing index×-0.485)+24.406)×0.353). Using a cut-off value of &gt;0.69 and Applying this index in phase II, it showed a mean value of -1.43 and 4.67 in the failure and success groups, respectively (P&lt;0.001). This new WI with cut-off value (&gt;0.69) had 88.5% sensitivity, 57.1% specificity, and a 77.5% accuracy to predict the success of weaning from MV. Conclusion We developed a new WI using formula including (semiquantitative Cough strength, PCO2 gradient mean airway pressure, PO2/FIO2 ratio, Rapid shallow breathing index) with accepted sensitivity and moderate specificity for predicting successful weaning from MV in ICU patients. Its use should be encouraged in the ICU setting, especially by less experienced ICU physicians.

Outcomes of Traumatic Brain Injury Patients Managed in a Non–Intensive Care Unit Setting

IntroductionThe management of traumatic brain injury (TBI) requires significant health-care resources. The modified Brain Injury Guidelines (mBIG) stratifies TBI patients by severity to help guide disposition and management. We sought to analyze the outcomes of TBI patients managed in a non–intensive care unit (ICU) setting after stratifying them using the mBIG criteria. MethodsA retrospective single-center study was performed on all adult patients who sustained blunt TBI from 2021 to 2022 and were managed in a non-ICU setting. Primary outcome was unplanned upgrade to the ICU. Secondary outcomes were need for neurosurgical intervention, unplanned intubation, mortality, and hospital length of stay. Patients were divided into cohorts of mBIG 1 & 2 versus mBIG 3. ResultsOf the 274 patients managed in a non-ICU setting, 119 (43.4%) met mBIG 3 criteria. The majority (76.5%) were managed in a step-down level of care. Nine patients required upgrade to the ICU, with only two upgraded for acute progression of their intracranial hemorrhage. Eight patients in mBIG 3 cohort required neurosurgical interventions, with only two related to progression of their intracranial hemorrhage and both over 24 h after admission. The remaining six patients had planned delayed neurosurgical intervention. Unplanned intubation occurred in three patients with only one related to a delayed progression of their TBI. Longer hospitalization and decreased survival were noted in mBIG 3 group. No differences in 30-d readmissions, stroke, venous thromboembolism events or seizures were found between the two groups. ConclusionsSelect patients with severe TBI may be considered for admission to step-down units with frequent neurologic exams in lieu of ICU level of care.

Sedation Vacations in Neurocritical Care: Friend or Foe?

To evaluate the role of sedation vacations in optimizing patient outcomes and enhancing the quality of care in neurological intensive care units (ICUs). We discuss the importance of sedation management in neurocritical care, considering recent research findings and clinical guidelines. Recent studies have highlighted the significance of sedation interruption protocols in improving patient outcomes in the ICU setting. Evidence suggests that daily sedation interruptions can reduce the duration of mechanical ventilation, ICU length of stay, and mortality rates. However, the implementation of these protocols requires careful consideration of patient-specific factors and a multidisciplinary approach. Sedation vacations play a critical role in neurocritical care by reducing mechanical ventilation duration, ICU stay length, and mortality rates. Despite the benefits, the presence of complications must be addressed to avoid adverse outcomes. Continued research is necessary to refine these strategies and improve guideline quality, ensuring safe and effective sedation management in critically ill neurological patients.

Assessing outcomes for patients receiving chemotherapy in the intensive care unit.

190 Background: Despite the low in-hospital mortality rate (9%) for Intensive Care Unit (ICU) stays in the US, outcomes for patients with cancer undergoing chemotherapy in the ICU remains unclear. We aimed to assess survival in patients with cancer (liquid, solid tumor and allogeneic stem cell transplant (alloSCT) receiving chemotherapy at our institution and identify patient characteristics for enhanced chemotherapy stewardship. Methods: We analyzed pharmacy data for patients with confirmed malignancies who received chemotherapy while admitted to the ICU at our institution from 1/1/21 to 12/31/23, excluding non-cancer patients and those on home oral regimens. Primary endpoints included 14-day, 30-day, and 90-day survival. Secondary analyses examined survival differences by age, ECOG performance status, and treatment history. Survival differences among patient subtypes were evaluated using Z-tests. Results: We identified 74 unique patients who received chemotherapy for cancer. 54 had liquid tumors, 5 were post-alloSCT, and 15 had solid tumors. The 90-day survival rate was 47% overall: 46% for the liquid tumor group, 80% for the alloSCT group, and 40% for the solid tumor group. When further stratified by tumor type, higher 90-day survival was variable; see Table 1 for details). In patients with lung cancer, 90-day survival was worse for non-small cell lung cancer without tumor mutations (0%, n=2) than those with targetable mutations (50%, n=4) or small cell lung cancer (33%, n=3). An ECOG performance status of 3-4 showed poorer 90-day survival in liquid tumors (23%) compared to ECOG 1-2 (74%; 95% CI 0.26 , 0.76). Solid tumors showed no significant ECOG differences (50%, 40%). Treatment-naive status showed higher 90-day survival in liquid tumors (59%) compared to treatment-resistant tumors (18%; 95% CI 0.17, 0.65), but not in solid tumors (50%, 20%). Conclusions: Our findings suggest generally unfavorable 90-day outcomes for ICU chemotherapy patients, especially those aged over 75, with multiple therapy lines, or ECOG status 3-4. Survival rates varied significantly across tumor types. We are currently broadening our study to include all solid tumor cancer patients. While this is a limited sample, this underscores the need for careful chemotherapy stewardship and advanced care planning in ICU settings. 90-day survival of solid and liquid tumor patients receiving chemotherapy in the ICU by select cancer diagnoses. Cancer Diagnosis n 90-Day Survival (%) AML 20 45 DLBCL 16 33 APML 6 83 Multiple myeloma 5 60 Other liquid tumors 1 11 45 Lung cancer 9 44 Germ cell tumor 2 100 Hormone-sensitive breast cancer 1 100 Other solid tumors 2 3 0 1 T-cell lymphoma, ALL, CNS lymphoma, Castleman, MPN, and plasmablastic lymphoma. Note, patients with allogenic stem cell transplants not included. 2 Carcinoma of unknown primary, colon adenocarcinoma, multiple solid tumors.

Using Machine Learning and Electronic Health Records to Identify Neuropsychiatric Risk Scores for Delirium in ICU and General Hospital Settings.

Delirium is a common and acute neuropsychiatric syndrome that requires timely intervention to prevent its associated morbidity and mortality. Yet, its diagnosis and symptoms are often overlooked due to its variable clinical presentation and fluctuating nature. Thus, in this study, we address the barriers to delirium diagnosis by utilizing a machine learning-based predictive algorithm for incident delirium that relies on archived electronic health records (EHRs) data. We used the Medical Information Mart for Intensive Care (MIMIC) database to create a detailed dataset for identifying delirium in intensive care unit (ICU) patients. Our approach involved training machine learning models on this dataset to pinpoint critical clinical features for delirium detection. These features were then refined and applied to non-ICU patients using EHRs from the American University of Beirut Medical Center (AUBMC). Our study assessed machine learning models like Extreme Gradient Boosting (XGBoost), Categorical Boosting (CatBoost), Classification and Regression Trees (CART), Random Forest (RF), Neural Oblivious Decision Ensembles (NODE), and Logistic Regression (LR), highlighting superior delirium detection in diverse clinical settings. The CatBoost model excelled in ICU environments with an F1 Score of 89.2%, while XGBoost performed best in general hospital settings with a 75.4% F1 Score. Interpretations using Tabular Local Interpretable Model-agnostic Explanations (LIME) revealed critical indicators such as prothrombin time and hematocrit levels, enhancing model transparency and clinical applicability. These clinical insights help differentiate the delirium predictors between ICU patients, who are often sensitive to various factors. The proposed predictive algorithm improves delirium detection rates and streamlines efficiency in hospital electronic systems, thereby enabling prompt interventions to prevent delirium progression and associated complications. The clinical indicators for delirium that we identified in general hospital settings and ICU can greatly help healthcare professionals identify potential causes of delirium and reduce misdiagnosis.

Emerging Colistin Resistance: Experience from an Intensive Care Unit of a Tertiary Care Center in India.

Resistance to the currently available classes of antimicrobials has emerged as a major health concern, with a growing number of isolated organisms demonstrating resistance to available antimicrobials. With the escalation of carbapenem resistance, the use of polymyxin B/E (colistin) has increased over the years, which has, in turn, contributed to the worldwide emergence of colistin resistance. The available data on colistin resistance from Southeast Asian nations, including India, is limited, especially in intensive care unit (ICU) settings. The primary objective of our study was to analyze the clinical profile of patients with polymyxin B/E (colistin) resistant positive cultures and to study their outcome in terms of length of ICU stay and outcome at discharge. This observational, single-center, prospective study was conducted in a 20-bed adult ICU serving both medical and surgical patients at a 1,600-bed tertiary care institute in northern India between Jan 2020 and Dec 2021. In this study, all adult patients, defined as individuals older than 18 years of age, admitted to our ICU with cultures detecting polymyxin B/E (colistin) resistant organisms were included as cases, and all adult ICU patients (age >18 years) with polymyxin B/E (colistin) sensitive cultures were taken as controls. Clinical, laboratory, and demographic parameters, along with ICU variables like severity of illness, length of ICU stay, mechanical ventilation, days of shock, renal replacement therapy (RRT), etc., and outcome at discharge were collected. Identification of resistance to colistin and minimal inhibitory concentration (MIC) of colistin is based on Clinical and Laboratory Standards Institute (CLSI) guidelines. For statistical analysis, Mann-Whitney U test, Fisher's exact test, and binary logistic regression were used. Twenty-eight cases and 55 controls (n = 83) were included for analysis. Abdominal [gastrointestinal (GI)] sepsis was the most common diagnosis (41.8%) at admission to the ICU, and Klebsiella pneumoniae was the most common species isolated in both the cases (96.4%) and the controls (50.9%). The most common site of isolation of Gram-negative bacteria in both cases and controls was blood (71.4 vs 74.5%), followed by deep-seated pus (21.4 vs 23.64%). The most common class of drugs to which the cases were sensitive was tetracyclines in 60.7%, followed by ceftazidime-avibactam. Factors such as prior exposure to colistin, exposure to monobactams in the ICU, and ICU stay in days were identified as independent predictors for colistin resistance. Overall mortality was not statistically different between cases and controls (p -value 0.38). Factors like prolonged ICU stay, exposure to monobactams in the ICU, and a history of prior exposure to colistin in the previous 90 days are independent predictors of colistin resistance. The most common organism to develop colistin resistance identified was K. pneumoniae, and the most common site of isolation was blood. There was no difference in mortality between colistin-resistant and colistin-sensitive patients.

Descriptive Analysis of Dexmedetomidine's Utility in a Palliative Care Unit at the End of Life.

Context: Pain and symptom management at the end of life (EoL) can pose unique challenges, particularly when symptoms are refractory to conventional methods. Dexmedetomidine, originally approved for sedation in ventilated patients, has been demonstrated to be beneficial in pain management and palliative care settings by functioning as an alpha-2 agonist. Methods: A retrospective review of inpatient palliative care unit (IPU) records from January 2020 to December 2023 was conducted. Twenty-five adult patients receiving continuous dexmedetomidine for refractory pain at the EoL were identified. These patients were further evaluated for concurrent opioid, benzodiazepine, and chlorpromazine usage. Results: Patients experienced predominantly cancer-related pain, and had a median infusion duration of 5 days. Dexmedetomidine's initial dosing differed between the intensive care unit (ICU) and IPU settings. There was a trend toward a decreased opioid requirement 24 hours after initiation. Patients transferred from the ICU showed a progressive increase in opioid use. Conclusion: This study contributes to understanding dexmedetomidine's role in managing refractory symptoms at the EoL in the palliative care setting.

Evaluating the current breadth of randomized control trials on cardiac arrest: A scoping review.

Despite the significant disease burden due to cardiac arrest, there is a relative paucity of randomized controlled trials (RCTs) to inform definitive management. We aimed to evaluate the current scope of cardiac arrest RCTs published between 2015 and 2022. We conducted a search in October 2023 of MEDLINE, Embase, and Web of Science for cardiac arrest RCTs. We included trials published between 2015 and 2022 enrolling human subjects suffering from non-traumatic cardiac arrest. Descriptive statistics were reported and the Mann Kendall test was used to evaluate for temporal trends in the number of trials published annually. We identified 1764 unique publications, 87 RCTs were included after title/abstract and full-text review. We found no significant increase in trials published annually (eight in 2015 and 16 in 2022, p=1.0). Geographic analysis of study centers found 31 countries represented; Denmark (n=13, 15%) and the United States (n=9, 10%) conducted the majority of trials. Nearly all trials included adults (n=84, 97%) and few included children (n=9, 10%). The majority of trials focused on out-of-hospital cardiac arrest (n=62, 71%). Thirty-eight (44%) trials used an intervention characterized as a process improvement; 28 (32%) interventions were characterized as a drug and 20 (23%) as a device. Interventions were implemented with similar frequency in the prehospital (33%) and intensive care unit (38%) setting, as well as similarly between the intra-arrest (53%) and post-arrest (46%) periods. Twenty (27%) trials selected a primary outcome of survival at ≥ 28 days. Publication of cardiac arrest RCTs remained constant between 2015 and 2022. We identified significant gaps including a lack of trials examining in-hospital cardiac arrest and pediatric patients.

Predictive Value of Combined Detection of Serum LGALS3BP and GDF-15 for the Prognosis of ICU Sepsis Patients.

This study aims to investigate the effectiveness of combining serum lectin galactoside-binding soluble 3 binding protein (LGALS3BP) with growth differentiation factor 15 (GDF-15) for predicting outcomes in sepsis patients in an intensive care unit (ICU) setting. The study involved 208 sepsis patients from the ICU of our hospital. These patients were categorized based on their 28-day survival outcomes into two groups: 166 in the survival group and 42 in the mortality group. The serum levels of LGALS3BP and GDF-15 were measured using the ELISA technique. Pearson and Spearman methods were utilized for correlation analysis. Factors affecting mortality in ICU sepsis patients were evaluated through multivariate logistic regression analysis. The efficacy of these biomarkers in prognosis prediction was assessed using receiver operating characteristic (ROC) curve analysis. The proportion of septic shock, APACHE II score, SOFA score, and serum LGALS3BP and GDF-15 levels in ICU sepsis patients in the death group were obviously higher than those in the survival group (P<0.05). The severity of ICU sepsis patients, APACHE II score, and SOFA score were obviously positively correlated with serum LGALS3BP and GDF-15 levels (P<0.05). LGALS3BP (OR: 95% CI=2.745:1.583~4.761) and GDF-15 (OR: 95% CI=2.639:1.423~4.893) were independent risk factors for death in ICU sepsis patients (P<0.05). The AUC of serum LGALS3BP and GDF-15 levels alone in predicting death in ICU sepsis patients was 0.859 and 0.854, obviously lower than the AUC of the combination, 0.943 (Z=2.704, 2.287, P<0.05). The AUC for predicting mortality in ICU sepsis patients using the APACHE II and SOFA scores were 0.832 and 0.842, respectively. The differences in comparison to the AUCs of LGALS3BP and GDF-15 were not statistically significant (P > 0.05). Serum levels of LGALS3BP and GDF-15 can both be used as predictive indicators for death in ICU sepsis patients, and their combined predictive efficacy is better.

Impact of daily music on comfort scores in preterm infants: a randomized controlled trial.

Premature birth heightens neurodevelopmental risks, theorized to partly stem from altered sensory inputs and disrupted sleep patterns. Modifying the acoustic milieu through music intervention (MI) offers promise to improve neonatal comfort, reduce sleep disturbances, and stabilize physiological parameters. This study explores the impact of non-pharmacological MI on these health indicators within the Neonatal Intensive Care Unit (NICU). A single-blinded RCT was conducted. Premature infants (34 > GA > 29 wks) were randomly assigned to either receive 8 min of daily MI or placebo for up to 15 days. Validated behavioral comfort scores were visually obtained by a blinded observer before and after intervention. Additionally, physiological signs (HR, RR, SatO2) were recorded. Differences between the groups were analyzed using χ² tests and t-tests. In total 56 preterm infants were included. After intervention, comfort levels increased significantly in the MI compared to placebo group (p = 0.000). Neonates receiving MI transitioned from wakefulness to a state of sleep significantly more compared to placebo (p = 0.002). Physiological parameters remained stable. This study adds to existing literature demonstrating that a music intervention in a NICU setting can enhance comfort and sleep of premature infants without adversely affecting physiological parameters. The study demonstrates that music interventions (MI) in a Neonatal Intensive Care Unit (NICU) setting can significantly increase comfort levels and positively impact the sleep of premature infants without negatively impacting their physiological parameters. This research increases evidence for non-pharmacological interventions, specifically music, as beneficial for the well-being of premature infants in NICU settings. It replicates and expands upon previous methodological designs, providing more robust evidence of MI's positive effects on this vulnerable population. The positive outcomes of music intervention could influence hospital policies by integrating non-pharmacological practices into standard neonatal care protocols to enhance developmental support for premature infants.

Identifying optimal serum 1,3-β-D-Glucan cut-off for diagnosing Pneumocystis Jirovecii Pneumonia in non-HIV patients in the intensive care unit

BackgroundSerum (1,3)-β-D-glucan (BDG) detection for diagnosis of Pneumocystis jirovecii pneumonia (PJP) in non-human immunodeficiency virus (HIV) immunocompromised patients lacks intensive care unit (ICU)-specific data. We aimed to assess its performance and determine the optimal cutoff for PJP in ICU population.MethodsThis retrospective study included critically ill non-HIV immunocompromised patients admitted to a medical ICU with suspected pneumonia, undergoing simultaneous microbiological testing for P. jirovecii on lower respiratory tract specimens and serum BDG. Confounders affecting BDG positivity were explored by multivariable logistic regression. Optimal cut-offs were derived from Youden’s index for the entire cohort and subgroups stratified by confounders. Diagnostic performance of serum BDG was estimated at different cutoffs.ResultsOf 400 patients included, 42% were diagnosed with PJP and 58.3% had positive serum BDG. Serum BDG’s area under the receiver operating characteristic curve was 0.90 (0.87–0.93). At manufacturer’s 150 pg/ml cut-off, serum BDG had high sensitivity and negative predictive value (94%), but low specificity and positive predictive value (67%). Confounders associated with a positive serum BDG in PJP diagnosis included IVIG infusion within 3 days (odds ratio [OR] 9.24; 95% confidence interval [CI] 4.09–20.88, p < 0.001), other invasive fungal infections (OR 4.46; 95% CI 2.10–9.49, p < 0.001) and gram-negative bacteremia (OR 29.02; 95% CI 9.03–93.23, p < 0.001). The application of optimal BDG cut-off values determined by Youden’s index (252 pg/ml, 390 pg/ml, and 202 pg/ml) specific for all patients and subgroups with or without confounders improved the specificity (79%, 74%, and 88%) and corresponding PPV (75%, 65%, and 85%), while maintaining reasonable sensitivity and NPV.ConclusionsTailoring serum BDG cutoff specific to PJP and incorporating consideration of confounders could enhance serum BDG’s diagnostic performance in the ICU settings.

Leveraging interpretable machine learning in intensive care

AbstractIn healthcare, especially within intensive care units (ICU), informed decision-making by medical professionals is crucial due to the complexity of medical data. Healthcare analytics seeks to support these decisions by generating accurate predictions through advanced machine learning (ML) models, such as boosted decision trees and random forests. While these models frequently exhibit accurate predictions across various medical tasks, they often lack interpretability. To address this challenge, researchers have developed interpretable ML models that balance accuracy and interpretability. In this study, we evaluate the performance gap between interpretable and black-box models in two healthcare prediction tasks, mortality and length-of-stay prediction in ICU settings. We focus specifically on the family of generalized additive models (GAMs) as powerful interpretable ML models. Our assessment uses the publicly available Medical Information Mart for Intensive Care dataset, and we analyze the models based on (i) predictive performance, (ii) the influence of compact feature sets (i.e., only few features) on predictive performance, and (iii) interpretability and consistency with medical knowledge. Our results show that interpretable models achieve competitive performance, with a minor decrease of 0.2–0.9 percentage points in area under the receiver operating characteristic relative to state-of-the-art black-box models, while preserving complete interpretability. This remains true even for parsimonious models that use only 2.2 % of patient features. Our study highlights the potential of interpretable models to improve decision-making in ICUs by providing medical professionals with easily understandable and verifiable predictions.

Risk factors for infection after carbapenem-resistant Acinetobacter baumannii colonization

PurposePredicting infection risk in carbapenem-resistant Acinetobacter baumannii (CRAB) colonized patients may help in improving timely appropriate antibiotic therapy. This study aims to explore risk factors for developing infections in hospitalized patients with previous CRAB colonization.MethodsWe performed an observational retrospective cohort study at ASST Sette Laghi-Varese Hospital between January 2020 and December 2022. All consecutive adult (> 18 years old) hospitalized patients with documented colonization by CRAB at any anatomical site or with CRAB infections preceded by CRAB colonization were included. Univariate and multivariate analyses were performed to investigate infection risk factors.ResultsOverall, 144 patients were included in the study: 104 colonized only and 40 infected patients. Colonization and infection rates significantly changed over the years (2020–2022, p < 0.001). The incidence of infections in CRAB carriers was 27.8% (40/144). Median time from colonization to infection was 4 days (IQR 1-8.5). Overall, inhospital mortality was 32.7% and 55.0% in colonized only and infected patients, respectively. At the multivariable logistic regression cardiovascular disease (OR 5.83, 95% CI 1.12–30.43, p = 0.037), COVID-19 (OR 3.72, 95% CI 1.16–11.91, p = 0.027) and intensive care unit (ICU) admission (OR 8.83, 95% CI 2.94–26.51, p < 0.001) were risk factors independently associated with cardiovascular disease CRAB infection after colonization.ConclusionsWe observed an increased infection risk in patients colonized with CRAB with cardiovascular disease, COVID-19 and admitted in ICU setting. Additional evidence is needed to identify predictors of infection in colonized patients.

A machine learning model for early candidemia prediction in the intensive care unit: Clinical application.

Candidemia often poses a diagnostic challenge due to the lack of specific clinical features, and delayed antifungal therapy can significantly increase mortality rates, particularly in the intensive care unit (ICU). This study aims to develop a machine learning predictive model for early candidemia diagnosis in ICU patients, leveraging their clinical information and findings. We conducted this study with a cohort of 334 patients admitted to the ICU unit at Ji Ning NO.1 people's hospital in China from Jan. 2015 to Dec. 2022. To ensure the model's reliability, we validated this model with an external group consisting of 77 patients from other sources. The candidemia to bacteremia ratio is 1:1. We collected relevant clinical procedures and eighteen key examinations or tests features to support the recursive feature elimination (RFE) algorithm. These features included total bilirubin, age, platelet count, hemoglobin, CVC, lymphocyte, Duration of stay in ICU and so on. To construct the candidemia diagnosis model, we employed random forest (RF) algorithm alongside other machine learning methods and conducted internal and external validation with training and testing sets allocated in a 7:3 ratio. The RF model demonstrated the highest area under the receiver operating characteristic (AUC) with values of 0.87 and 0.83 for internal and external validation, respectively. To evaluate the importance of features in predicting candidemia, Shapley additive explanation (SHAP) values were calculated and results revealed that total bilirubin and age were the most important factors in the prediction model. This advancement in candidemia prediction holds significant promise for early intervention and improved patient outcomes in the ICU setting, where timely diagnosis is of paramount crucial.

Alarm Management in Provisional COVID-19 Intensive Care Units: Retrospective Analysis and Recommendations for Future Pandemics.

In response to the high patient admission rates during the COVID-19 pandemic, provisional intensive care units (ICUs) were set up, equipped with temporary monitoring and alarm systems. We sought to find out whether the provisional ICU setting led to a higher alarm burden and more staff with alarm fatigue. We aimed to compare alarm situations between provisional COVID-19 ICUs and non-COVID-19 ICUs during the second COVID-19 wave in Berlin, Germany. The study focused on measuring alarms per bed per day, identifying medical devices with higher alarm frequencies in COVID-19 settings, evaluating the median duration of alarms in both types of ICUs, and assessing the level of alarm fatigue experienced by health care staff. Our approach involved a comparative analysis of alarm data from 2 provisional COVID-19 ICUs and 2 standard non-COVID-19 ICUs. Through interviews with medical experts, we formulated hypotheses about potential differences in alarm load, alarm duration, alarm types, and staff alarm fatigue between the 2 ICU types. We analyzed alarm log data from the patient monitoring systems of all 4 ICUs to inferentially assess the differences. In addition, we assessed staff alarm fatigue with a questionnaire, aiming to comprehensively understand the impact of the alarm situation on health care personnel. COVID-19 ICUs had significantly more alarms per bed per day than non-COVID-19 ICUs (P<.001), and the majority of the staff lacked experience with the alarm system. The overall median alarm duration was similar in both ICU types. We found no COVID-19-specific alarm patterns. The alarm fatigue questionnaire results suggest that staff in both types of ICUs experienced alarm fatigue. However, physicians and nurses who were working in COVID-19 ICUs reported a significantly higher level of alarm fatigue (P=.04). Staff in COVID-19 ICUs were exposed to a higher alarm load, and the majority lacked experience with alarm management and the alarm system. We recommend training and educating ICU staff in alarm management, emphasizing the importance of alarm management training as part of the preparations for future pandemics. However, the limitations of our study design and the specific pandemic conditions warrant further studies to confirm these findings and to explore effective alarm management strategies in different ICU settings.

Defining minimum workforce standards for intensive care physiotherapy in Australia and New Zealand: A Delphi study

BackgroundIntensive care staffing guidelines provide recommendations for the safe and effective delivery of health care while recognising professional requirements of the workforce. To guide recommendations for physiotherapy staffing guidelines, profession-specific consultation is needed. ObjectivesThe objective of this study was to develop consensus-based recommendations for minimum workforce standards for physiotherapy in intensive care. MethodsA Delphi survey process was conducted involving physiotherapists from Australia and New Zealand. ResultsThe panel consisted of 65 physiotherapists in the first round and 60 in the second round (92% retention). Respondents were from both Australia (49, 76%) and New Zealand (16, 24%) who had been physiotherapists for an average of 18.8 ± 9.0 years and were primarily senior intensive care physiotherapists (44, 68%). Respondents had experience across level 3 (50, 77%), level 2 (18, 28%), and level 1 (5, 8%) adult intensive care units (ICUs), adult high-dependency units (27, 42%), and paediatric intensive care (6, 9%). A total of 42 statements were presented, with 37 reaching consensus after two rounds. After two rounds, consensus was achieved for a minimum staffing ratio in paediatric ICUs of one physiotherapist per six (1:6) beds. For adult ICUs, use of the median value of the participant's responses was supported to establish minimum staffing ratios of 1:8, 1:7, 1:6, and 1:8 for levels 1, 2, and 3 ICUs and high-dependency units, respectively. The requirement for an additional allocation for senior physiotherapist staffing for each ICU level was also established. Statements that also gained consensus included recommendations for access to on-call and weekend services for all ICU settings and the consideration of evening shifts specifically for level 3 and paediatric ICUs. ConclusionsRecommendations for minimum staffing for physiotherapy in intensive care settings were achieved and supported requirements for clinical service delivery, supervision, and training.

The Influence of Ethical Nursing Competence and Positive Nursing Organizational Culture on Person-Centered Care in Intensive Care Unit Nurses: A Cross-Sectional Survey

Purpose: The purpose of this study was to investigate the impact of ethical nursing competence and positive nursing organizational culture on person-centered care among intensive care unit (ICU) nurses.Methods: A cross-sectional research design was adopted for this study. The participants were 156 ICU nurses working at three hospitals in Seoul City and Gyeonggi-do. The data were collected from September 8 to September 30, 2023. The questionnaire consisted of tools for general characteristics, person-centered care, ethical nursing competence, and positive nursing organizational culture.Results: Participants had a mean age of 28.90±5.38 years, and the mean score for person-centered care was 3.65±0.46. Person-centered care was influenced by ethical nursing competence (β=.60, p&lt;.001) and positive nursing organizational culture (β=.21, p=.004). The explanatory power was 32.4% (F=13.36, p&lt;.001, Adj. R2=.324).Conclusion: Based on the findings of this study, to improve the performance of person-centered nursing by ICU nurses, it is necessary to provide ongoing, systematic education focusing on ethical nursing. This educational initiative should be designed to address the specific ethical challenges encountered in ICU settings. In addition, strengthening managers’ active leadership and promoting a positive nursing organizational culture can be expected to improve person-centered nursing performance.