ABSTRACT
Background: We assessed 2 versions of the large language model (LLM) ChatGPT-versions 3.5 and 4.0-in generating appropriate, consistent, and readable recommendations on core critical care topics. Research Question: How do successive large language models compare in terms of generating appropriate, consistent, and readable recommendations on core critical care topics? Design and Methods: A set of 50 LLM-generated responses to clinical questions were evaluated by 2 independent intensivists based on a 5-point Likert scale for appropriateness, consistency, and readability. Results: ChatGPT 4.0 showed significantly higher median appropriateness scores compared to ChatGPT 3.5 (4.0 vs 3.0, P < .001). However, there was no significant difference in consistency between the 2 versions (40% vs 28%, P = 0.291). Readability, assessed by the Flesch-Kincaid Grade Level, was also not significantly different between the 2 models (14.3 vs 14.4, P = 0.93). Interpretation: Both models produced "hallucinations"-misinformation delivered with high confidence-which highlights the risk of relying on these tools without domain expertise. Despite potential for clinical application, both models lacked consistency producing different results when asked the same question multiple times. The study underscores the need for clinicians to understand the strengths and limitations of LLMs for safe and effective implementation in critical care settings. Registration: https://osf.io/8chj7/.
ABSTRACT
Importance: Withdrawal of life-sustaining therapy (WLST) decisions for critically injured trauma patients are complicated and multifactorial, with potential for patients' insurance status to affect decision-making. Objectives: To determine if patient insurance type (private insurance, Medicaid, and uninsured) is associated with time to WLST in critically injured adults cared for at US trauma centers. Design, Setting, and Participants: This retrospective registry-based cohort study included reported data from level I and level II trauma centers in the US that participated in the American College of Surgeons Trauma Quality Improvement Program (TQIP) registry. Participants included adult trauma patients who were injured between January 1, 2017, and December 31, 2020, and required an intensive care unit stay. Patients were excluded if they died on arrival or in the emergency department or had a preexisting do not resuscitate directive. Analyses were performed on December 12, 2023. Exposures: Insurance type (private insurance, Medicaid, uninsured). Main Outcomes and Measures: An adjusted time-to-event analysis for association between insurance status and time to WLST was performed, with analyses accounting for clustering by hospital. Results: This study included 307â¯731 patients, of whom 160â¯809 (52.3%) had private insurance, 88â¯233 (28.6%) had Medicaid, and 58â¯689 (19.1%) were uninsured. The mean (SD) age was 40.2 (14.1) years, 232â¯994 (75.7%) were male, 59â¯551 (19.4%) were African American or Black patients, and 201â¯012 (65.3%) were White patients. In total, 12â¯962 patients (4.2%) underwent WLST during their admission. Patients who are uninsured were significantly more likely to undergo earlier WLST compared with those with private insurance (HR, 1.54; 95% CI, 1.46-1.62) and Medicaid (HR, 1.47; 95% CI, 1.39-1.55). This finding was robust to sensitivity analysis excluding patients who died within 48 hours of presentation and after accounting for nonwithdrawal death as a competing risk. Conclusions and Relevance: In this cohort study of US adult trauma patients who were critically injured, patients who were uninsured underwent earlier WLST compared with those with private or Medicaid insurance. Based on our findings, patient's ability to pay was may be associated with a shift in decision-making for WLST, suggesting the influence of socioeconomics on patient outcomes.
Subject(s)
Insurance Coverage , Withholding Treatment , Wounds and Injuries , Humans , Male , Female , Middle Aged , Withholding Treatment/statistics & numerical data , Adult , Retrospective Studies , Wounds and Injuries/therapy , United States , Insurance Coverage/statistics & numerical data , Medicaid/statistics & numerical data , Insurance, Health/statistics & numerical data , Registries , Trauma Centers/statistics & numerical data , Medically Uninsured/statistics & numerical data , Critical Illness/therapy , Life Support Care/statistics & numerical data , AgedABSTRACT
Traumatic brain injury (TBI) is a leading cause of mortality and morbidity amongst trauma patients. Its treatment is focused on minimizing progression to secondary injury. Administration of propranolol for TBI maydecrease mortality and improve functional outcomes. However, it is our sense that its use has not been universally adopted due to low certainty evidence. The literature was reviewed to explore the mechanism of propranolol as a therapeutic intervention in TBI to guide future clinical investigations. Medline, Embase, and Scopus were searched for studies that investigated the effect of propranolol on TBI in animal models from inception until June 6, 2023. All routes of administration for propranolol were included and the following outcomes were evaluated: cognitive functions, physiological and immunological responses. Screening and data extraction were done independently and in duplicate. The risk of bias for each individual study was assessed using the SYCLE's risk of bias tool for animal studies. Three hundred twenty-three citations were identified and 14 studies met our eligibility criteria. The data suggests that propranolol may improve post-TBI cognitive and motor function by increasing cerebral perfusion, reducing neural injury, cell death, leukocyte mobilization and p-tau accumulation in animal models. Propranolol may also attenuate TBI-induced immunodeficiency and provide cardioprotective effects by mitigating damage to the myocardium caused by oxidative stress. This systematic review demonstrates that propranolol may be therapeutic in TBI by improving cognitive and motor function while regulating T lymphocyte response and levels of myocardial reactive oxygen species. Oral or intravenous injection of propranolol following TBI is associated with improved cerebral perfusion, reduced neuroinflammation, reduced immunodeficiency, and cardio-neuroprotection in preclinical studies.
Subject(s)
Brain Injuries, Traumatic , Propranolol , Propranolol/pharmacology , Propranolol/therapeutic use , Animals , Brain Injuries, Traumatic/drug therapy , Neuroprotective Agents/therapeutic use , Neuroprotective Agents/pharmacology , Humans , Disease Models, Animal , Drug Evaluation, Preclinical , Adrenergic beta-Antagonists/pharmacology , Adrenergic beta-Antagonists/therapeutic useABSTRACT
Objective: To evaluate transperineal laser ablation (TPLA) with Echolaser® (Echolaser® TPLA, Elesta S.p.A., Calenzano, Italy) as a treatment for benign prostatic hyperplasia (BPH) and prostate cancer (PCa) using the Delphi consensus method. Methods: Italian and international experts on BPH and PCa participated in a collaborative consensus project. During two rounds, they expressed their opinions on Echolaser® TPLA for the treatment of BPH and PCa answering online questionnaires on indications, methodology, and potential complications of this technology. Level of agreement or disagreement to reach consensus was set at 75%. If the consensus was not achieved, questions were modified after each round. A final round was performed during an online meeting, in which results were discussed and finalized. Results: Thirty-two out of forty invited experts participated and consensus was reached on all topics. Agreement was achieved on recommending Echolaser® TPLA as a treatment of BPH in patients with ample range of prostate volume, from <40 mL (80%) to >80 mL (80%), comorbidities (100%), antiplatelet or anticoagulant treatment (96%), indwelling catheter (77%), and strong will of preserving ejaculatory function (100%). Majority of respondents agreed that Echolaser® TPLA is a potential option for the treatment of localized PCa (78%) and recommended it for low-risk PCa (90%). During the final round, experts concluded that it can be used for intermediate-risk PCa and it should be proposed as an effective alternative to radical prostatectomy for patients with strong will of avoiding urinary incontinence and sexual dysfunction. Almost all participants agreed that the transperineal approach of this organ-sparing technique is safer than transrectal and transurethral approaches typical of other techniques (97% of agreement among experts). Pre-procedural assessment, technical aspects, post-procedural catheterization, pharmacological therapy, and expected outcomes were discussed, leading to statements and recommendations. Conclusion: Echolaser® TPLA is a safe and effective procedure that treats BPH and localized PCa with satisfactory functional and sexual outcomes.
ABSTRACT
INTRODUCTION: Blunt cerebrovascular injury (BCVI) occurs in 1-3% of blunt traumas and is associated with stroke, disability, and mortality if unrecognized and untreated. Early detection and treatment are imperative to reduce the risk of stroke, however, there is significant variation amongst centers and trauma care providers in the specific medical management strategy used. This study compares antiplatelets vs. anticoagulants to determine BCVI-related stroke risk and bleeding complications to better understand the efficacy and safety of various treatment strategies. METHODS: A systematic review of MEDLINE, Embase, and Cochrane CENTRAL databases was conducted with the assistance of a medical librarian. The search was supplemented with manual review of the literature. Included studies reported treatment-stratified risk of stroke following BCVI. All studies were screened independently by two reviewers, and data was extracted in duplicate. Meta-analysis was conducted using pooled estimates of odds ratios (OR) with a random-effects model using Mantel-Haenszel methods. RESULTS: A total of 3315 studies screened yielded 39 studies for inclusion, evaluating 6552 patients (range 8 - 920 per study) with a total of 7643 BCVI. Stroke rates ranged from 0% to 32.8%. Amongst studies included in the meta-analysis, there were a total of 405 strokes, with 144 (35.5%) occurring on therapy, for a total stroke rate of 4.5 %. Meta-analysis showed that stroke rate after BCVI was lower for patients treated with antiplatelets vs. anticoagulants (OR 0.57; 95% CI 0.33-0.96, p = 0.04); when evaluating only the 9 studies specifically comparing ASA to heparin, the stroke rate was similar between groups (OR 0.43; 95% CI 0.15-1.20, p = 0.11). Eleven studies evaluated bleeding complications and demonstrated lower risk of bleeding with antiplatelets vs. anticoagulants (OR 0.29; 95% CI 0.13-0.63, p = 0.002); 5 studies evaluating risk of bleeding complications with ASA vs. heparin showed lower rates of bleeding complications with ASA (OR 0.16; 95% CI 0.04-0.58, p = 0.005). CONCLUSIONS: Treatment of patients with BCVI with antiplatelets is associated with lower risks of stroke and bleeding complications compared to treatment with anticoagulants. Use of ASA vs. heparin specifically was not associated with differences in stroke risk, however, patients treated with ASA had fewer bleeding complications. Based on this evidence, antiplatelets should be the preferred treatment strategy for patients with BCVI.
Subject(s)
Cerebrovascular Trauma , Stroke , Wounds, Nonpenetrating , Humans , Anticoagulants/adverse effects , Anticoagulants/therapeutic use , Cerebrovascular Trauma/complications , Heparin/adverse effects , Heparin/therapeutic use , Retrospective Studies , Stroke/etiology , Wounds, Nonpenetrating/therapyABSTRACT
Predicting inpatient length of stay (LoS) is important for hospitals aiming to improve service efficiency and enhance management capabilities. Patient medical records are strongly associated with LoS. However, due to diverse modalities, heterogeneity, and complexity of data, it becomes challenging to effectively leverage these heterogeneous data to put forth a predictive model that can accurately predict LoS. To address the challenge, this study aims to establish a novel data-fusion model, termed as DF-Mdl, to integrate heterogeneous clinical data for predicting the LoS of inpatients between hospital discharge and admission. Multi-modal data such as demographic data, clinical notes, laboratory test results, and medical images are utilized in our proposed methodology with individual "basic" sub-models separately applied to each different data modality. Specifically, a convolutional neural network (CNN) model, which we termed CRXMDL, is designed for chest X-ray (CXR) image data, two long short-term memory networks are used to extract features from long text data, and a novel attention-embedded 1D convolutional neural network is developed to extract useful information from numerical data. Finally, these basic models are integrated to form a new data-fusion model (DF-Mdl) for inpatient LoS prediction. The proposed method attains the best R2 and EVAR values of 0.6039 and 0.6042 among competitors for the LoS prediction on the Medical Information Mart for Intensive Care (MIMIC)-IV test dataset. Empirical evidence suggests better performance compared with other state-of-the-art (SOTA) methods, which demonstrates the effectiveness and feasibility of the proposed approach.
Subject(s)
Inpatients , Learning , Humans , Length of Stay , Hospitalization , Critical CareABSTRACT
BACKGROUND: Resuscitative transesophageal echocardiography (TEE) is an emerging POCUS modality that can be used to guide trauma resuscitation. METHODS: Trauma patients who underwent TEE within 24 h of admission from 2013 to 2022 were prospectively identified. We retrospectively analyzed resuscitative TEE reports and patient charts in duplicate. RESULTS: 29 providers performed TEE for 54 acute trauma patients. 28 (52%) died in hospital; 33 (61%) required operative intervention (<24 h). Median injury severity score was 29 [IQR 22-43]. The most common indications for TEE were hemodynamic instability (34, 63%), inadequate windows for transthoracic echocardiography (14, 26%) and cardiac arrest (11, 20%). There were no identified complications. A new diagnosis was made in 31 (57%) cases: most commonly right ventricular dysfunction (10, 19%), pericardial effusion (9, 17%), and hypovolemia (6, 11%). TEE ruled out major cardiac injury in 83% of cases. TEE changed resuscitative strategy, in 17 (32%) patients, diagnostic imaging approach in 6 (11%) patients, procedural or operative approach in 5 (9%) patients and disposition from the trauma bay in 4 (7%) patients. CONCLUSION: Resuscitative TEE during acute trauma care has an additional diagnostic yield to existing diagnostic pathways and may impact definitive management for some patients in the trauma bay.