Citations

Abstract Effective communication in the intensive care unit (ICU) is essential, particularly for patients with non-English language preference, yet timely access to professional interpreters remains limited. While artificial intelligence (AI)-based translation tools have been explored in outpatient and nonacute care settings, studies evaluating their use in acute care, environments such as the ICU remain limited. To address this gap, we developed AI-TransLATE (AI-enhanced Transition to Language-Agnostic Transcultural Engagement), a speech-based translation tool designed for multilingual communication in critical care settings. This study aimed to assess the interpretation quality of AI-TransLATE across four languages—Spanish, Chinese, Arabic, and Turkish—using scripted ICU scenarios. We created ICU communication scripts and recorded bilingual research team members simulating clinical interactions. Two independent bilingual evaluators assessed interpretation quality using a 5-point Likert scale across fluency, adequacy, meaning preservation, and severity of errors. Clarity and cultural appropriateness were also rated. Percentage agreement was used to assess interrater agreement. AI-TransLATE achieved acceptable composite scores (≥16/20) across all languages. Spanish and Turkish performed consistently well; Chinese and Arabic showed variability due to omissions and terminology errors. AI-TransLATE shows promise as a clinical communication tool, but further evaluation in real-world, unscripted ICU settings is needed.

Acute respiratory distress syndrome (ARDS) is a common and highly heterogeneous condition in the critically ill. The association between hyper- and hypo-inflammatory subphenotypes and clinical outcomes has generated significant interest in precise ARDS management. The value of identifying biomarkers to guide treatment and enrollment in future ARDS trials is undisputable. We describe multiple factors complicating the search for subphenotypes and their treatable traits. The observed heterogeneity seen in the clinical course of ARDS is dynamic and influenced by factors beyond lung pathophysiology, including variations in the delivery of best critical care practices, patient comorbidities, and functional status, and patient or family preferences. Current subphenotype definitions lack strong biological plausibility and without clear evidence of benefit from targeted treatments, their use in clinical practice is currently unwarranted.

Abstract Introduction: The variety of pulmonary function tests (PFTs) offered in the pulmonary function lab (PFL) can confuse providers unfamiliar with the clinical role of the individual components of the PFT, potentially leading to repeat tests, diagnostic delays, patient burden, and reduced lab efficiency. We hypothesized that a symptom- and diagnosis-based ordering set will reduce the need for repeat testing visits. Methods: Baseline data (June 2022-June 2023) were first collected to establish PFL performance benchmarks and identify common reasons for incomplete or repeat PFL visits. We then implemented an electronic order set reflecting the clinical need of the PFT, based on symptoms, diagnoses, suspected diagnoses, or other common clinical indication in addition to a physician-supervised, results-driven protocol to allow de-escalation of testing when appropriate The pre-implementation data was compared to the post-implementation data (July 2023-June 2024), evaluating for its impact on PFL access, utilization of the specific PFT components (e.g. methacholine challenge test, MCT), and on diagnoses. Results: During the study period 52,292 PFTs were performed during the pre-implementation phase compared to 58,310 tests in the post-implementation phase. Following intervention, the need for a second PFL visit for the same clinical indication decreased from 26% to 13% (p<0.0001). The same day PFL capacity increased by 3-4 patients per day. 80% of primary care providers reported the new order design was more intuitive and easier to use. Among the specific components of the PFT, there was a significant increase in the number MCTs being performed, from 2.5% to 4%, (p<0.0001), with a change rate ranging between 26 – 81% / month. In a subset analysis of 57 consecutive patients having MCTs, the primary indications were evaluation for cough or wheezing (57.9%), initial asthma evaluation (22.8%), and specifically requested by the ordering provider (19.3%). MCT was not performed in 23% due to a low baseline lung function, difficulty with repeatable efforts, recent bronchodilator administration, or due to complicating comorbidities. Of the 39 tests completed, 89% were deemed appropriate, with 32% having positive results and 42% ultimately diagnosed with asthma. Conclusions: A symptom and/or diagnosis guided ordering set for pulmonary function testing appears to improve the efficiency of the PFL, while improving provider experience, PFL access, as well as on patient burden and diagnoses. Further research is needed to determine potential benefits for provider decision-making and patient outcomes.

Abstract Rationale: Immune checkpoint inhibitors (ICIs) have revolutionized the treatment of cancer. They are also associated with life-threatening immune-related adverse events (irAEs), including the rare, poorly understood, and potentially fatal triad of myasthenia gravis, myocarditis, and myositis(triple M syndrome). We aimed to describe the management, intensive care unit (ICU)utilization, and outcomes of patients with irAEs with a specific focus on triple M syndrome. Methods: We conducted a retrospective chart review of patients with irAEs admitted to three quaternary care hospitals between December 2018 and 2023. We aimed to collect data on demographics, cancer type, treatment, ICU resource utilization, and outcomes including mortality. Results: Over this 5-year period, 140 patients were admitted for irAEs. Forty-nine percent were male, median age was 73 (IQR 64-80) years, 7% were of African American or Asian descent, and 32%required ICU admission. Median ICU and hospital length of stay (LOS) were 1.5 (IQR 0.7-2.9)and 7.2 (IQR 4.1-13.4) days, respectively. Among ICU patients, 14/140 (10%) had myositis,22/140 (16%) myocarditis, and 9/140 (6.4%) myasthenia gravis. Triple-M syndrome was diagnosed in 7/140 (5%) patients, including 3 with metastatic melanoma, 2 with genitourinary malignancies, 1 with metastatic lung and 1 with colon cancer. Four had received pembrolizumab and 3 nivolumab. Median ICU and hospital LOS were 3 (IQR 1-6) and 10 (IQR 8-21) days in this group. Three patients required mechanical ventilation, one required tracheostomy, and one required vasopressor support. Triple-M management included corticosteroids, IVIG, pyridostigmine, mycophenolate mofetil, and rituximab. Patients with triple-M syndrome were significantly older than other patients with ICI toxicity (P <0.009) and had higher mortality (57%vs 25%, p = 0.07). Patients discharged from the ICU had a median survival of 67 (IQR 26-201)days. Conclusions: Patients with irAEs frequently require ICU level of support with significant morbidity and mortality, especially with triple M syndrome. Clinicians should always consider the possibility of triple-M syndrome in patients with multiple irAEs as symptoms can be subtle and easily overlooked. Corticosteroids remain the primary treatment of irAEs and can be combined with other immunosuppressive agents such as infliximab, anti-thymocyte globulin, or pyridostigmine. Unfortunately, mortality due to triple-M syndrome remains high underscoring the importance of early recognition and prompt multidisciplinary management of these toxicities.

To identify medical phrases utilized by the critical care team that may have an unintended impact on the critically ill patient, we administered an anonymous survey to multi-professional critical care team members. We elicited examples of imprecise language that may have a negative emotional impact on the critically ill. Of the 1600 providers surveyed, 265 offered 1379 examples (912 unique) which were clustered into 5 categories. Medical jargon (eg, “riding the vent”) was most prevalent (n = 549). There were 217 negative suggestions (eg, “you will feel a stick and a burn”). Hyperboles (eg, “black cloud”) were common (n = 198) while homonyms (ie “he fibbed”) accounted for 150 examples. Phrases such as “code brown in there” were categorized as metonyms (n = 144). 121 metaphors/similes (eg, “rearranging deck chairs on the Titanic”) were provided. Phrases that have the potential to negatively impact critically ill patient perceptions are commonplace in critical care practice. Whether these everyday communication habits lead to an unintended nocebo effect on mental health outcomes of the critically ill deserves further study.