Sen, A

Emerging evidence links veno-venous extracorporeal membrane oxygenation (VV-ECMO) to right ventricular (RV) dysfunction, but the relative contribution of ECMO versus underlying disease progression remains unclear. We conducted a multicenter, retrospective, propensity-matched study including 408 intubated acute respiratory distress syndrome (ARDS) patients who were admitted between January 2020 and October 2024 in a quaternary referral hospital network within the United States and underwent at least two transthoracic echocardiograms (TTEs) after intubation. Using 1:2 propensity score matching, we adjusted for baseline comorbidities, baseline hemoglobin, lactate, and worst PaO 2 /FiO 2 ratio (P/F). Longitudinal echocardiographic changes were analyzed with a cumulative link mixed model (CLMM), evaluating RV size and function as ordinal outcomes. Within 408 patients (112 received ECMO support), propensity matching identified 55 ECMO and 110 non-ECMO patients with balanced baseline characteristics. Among 469 serial echocardiograms, RV size and function worsened more frequently in ECMO patients. In longitudinal modeling, ECMO use alone was not associated with RV dilation ( p = 0.23) or dysfunction ( p = 0.43). However, longer ECMO duration was associated with RV dilation (OR: 1.04 per day, p = 0.004) and dysfunction (odds ratio [OR]: 1.04 per day, p < 0.001). Higher ECMO circuit flow had a trend toward progressive RV dilation and dysfunction but did not reach statistical significance. These findings highlight the need for vigilant RV monitoring during ECMO support.

Abstract Background Stroke is a time-sensitive neurological emergency in which early EMS activation and presentation to definitive care are cornerstones of effective therapy. Large language models (LLMs) are increasingly consulted by the public for medical advice, but the veracity of the guidance provided by commercially available models responding to potential stroke symptoms is not well understood. Methods We performed a cross-model benchmarking study comparing the triage choices of three frontier LLMs (Claude Sonnet 4.6, GPT-4o, and Llama 3.3-70b-versatile) on first-person vignettes describing a unilateral arm symptom on waking, across 10 symptom descriptors, and two clinical phases (before and after a partially reassuring self-examination), with or without a clinical distractor (n=50 per condition). Results Claude sought emergency care most often, Llama least, and GPT-4o in between, diverging most sharply in the post-examination phase where Claude called 911 in 100% of runs, Llama called for non-emergency help in 100%, and GPT-4o was symptom-dependent. A distractor shifted behavior away from emergency care in almost all conditions: calling 911 fell from 37.9% to 14.6% and waiting rose from 0% to 45.9% in the post-examination vignette. Responses were also sensitive to symptom word: weak, limp, heavy, and clumsy generated higher alarm, whereas numb, tingly, odd, strange, and weird generated less urgent responses. Conclusions The increasing use of LLMs for medical advice has significant public health implications. Commercially available LLMs show significant model-to-model variability and framing sensitivity when confronted with potential stroke symptoms, including under-recognition of canonical CDC warning descriptors, underscoring the need for systematic benchmarking as these tools become de facto first points of contact for patients experiencing neurological emergencies.

Abstract Background Diuretic resistance (DR) is the failure to achieve adequate natriuresis despite high diuretic doses, reflecting a state of low diuretic efficiency (DE). In heart failure (HF) and general intensive care cohorts, low DE and the presence of DR have been associated with adverse outcomes, including increased risk of rehospitalization and mortality (1-3). Quantitative data, however, remains limited in patients admitted to the cardiac intensive care unit (CICU) – a population characterized by cardiac critical illnesses and substantial cardiorenal interactions. Purpose The aim of this study is to investigate and characterize the association between DR and associated mortality outcomes in CICU patients. Methods This retrospective cohort study included adult CICU patients (2007-2018) who received loop diuretics on day 1 of admission. Patients with a stay &lt;1 day or oliguria [&lt;400mL urine output (UOP) total] were excluded. DR was defined as &lt;500mL of UOP per 40mg oral furosemide equivalent (FSE). Acute kidney injury (AKI) was defined per KDIGO guidelines. Primary outcome was all-cause in-hospital mortality (including CICU deaths), and secondary outcome was all-cause 1-year mortality. In-hospital mortality predictors were evaluated using logistic regression and 1-year mortality predictors by Cox proportional-hazards models. 1-year survival was assessed using Kaplan-Meier curves. Multivariable models were adjusted for known confounders. Results 3,229 total patients were included. Median age was 71.2 (61.2, 80.7) years and 41.3% were female. Admission diagnoses included HF exacerbation (79.9%), acute coronary syndrome (34.5%), and respiratory failure (38.9%), among others. 1,098 (34.0%) patients met DR criterion; they had higher severity of illness scores, more comorbidities (including HF, diabetes mellitus, lung disease, and chronic kidney disease, p&lt;0.001), and greater laboratory derangements. On admission, 23.6% patients had AKI, of which 54.3% patients met DR criterion. During hospitalization, 56.0% patients had AKI; those with DR had higher prevalence and severity of AKI (unadjusted OR 3.55 [3.02-4.17], p&lt;0.001). Overall in-hospital mortality was 9.8% (including 5.2% in the CICU). In patients with DR, in-hospital mortality was 17.0% (9.7% in the CICU). Notably, 59.4% of all in-hospital deaths occurred in patients with DR. DR was independently associated with higher in-hospital (adjusted OR 2.00 [1.51-2.64], p&lt;0.001) and 1-year (adjusted HR 1.64 [1.44-1.88], p&lt;0.001) mortality. Patients with DR had higher mortality in each AKI stage (Figure 1), and patients with concomitant DR and AKI had the highest risk of 1-year mortality (Figure 2). Conclusion DR, defined by our easily applicable threshold, emerged as a powerful, independent predictor of adverse outcomes and was associated with substantially higher mortality, demonstrating potential as a valuable risk stratifying tool.