Explore the vast range of titles available.

Background Large language models are increasingly evaluated for use in healthcare. However, concerns about their impact on disparities persist. This study reviews current research on demographic biases in large language models to identify prevalent bias types, assess measurement methods, and evaluate mitigation strategies. Methods We conducted a systematic review, searching publications from January 2018 to July 2024 across five databases. We included peer-reviewed studies evaluating demographic biases in large language models, focusing on gender, race, ethnicity, age, and other factors. Study quality was assessed using the Joanna Briggs Institute Critical Appraisal Tools. Results Our review included 24 studies. Of these, 22 (91.7%) identified biases. Gender bias was the most prevalent, reported in 15 of 16 studies (93.7%). Racial or ethnic biases were observed in 10 of 11 studies (90.9%). Only two studies found minimal or no bias in certain contexts. Mitigation strategies mainly included prompt engineering, with varying effectiveness. However, these findings are tempered by a potential publication bias, as studies with negative results are less frequently published. Conclusion Biases are observed in large language models across various medical domains. While bias detection is improving, effective mitigation strategies are still developing. As LLMs increasingly influence critical decisions, addressing these biases and their resultant disparities is essential for ensuring fair artificial intelligence systems. Future research should focus on a wider range of demographic factors, intersectional analyses, and non-Western cultural contexts. Graphic Abstract

Over the last 20 years, numerous research articles and clinical guidelines aimed at optimizing resource utilization for emergency department (ED) patients presenting with syncope have been published. We hypothesized that there would be temporal trends in syncope-related ED visits and associated trends in imaging, hospital admissions, and diagnostic frequencies. The ED component of National Hospital Ambulatory Medical Care Survey was analyzed from 2001 through 2010, comprising more than 358000 visits (representing an estimated 1.18 billion visits nationally). We selected ED visits with a reason for visit of syncope or fainting and calculated nationally representative weighted estimates for prevalence of such visits and associated rates of advanced imaging utilization and admission. For admitted patients from 2005 to 2010, the most frequent hospital discharge diagnoses were tabulated. During the study period, there were more than 3500 actual ED visits (representing 11.9 million visits nationally) related to syncope, representing roughly 1% of all ED visits. Admission rates for syncope patients ranged from 27% to 35% and showed no significant downward trend (P = .1). Advanced imaging rates increased from about 21% to 45% and showed a significant upward trend (P < .001). For admitted patients, the most common hospital discharge diagnosis was the symptomatic diagnosis of “syncope and collapse” (36.4%). Despite substantial efforts by medical researchers and professional societies, resource utilization associated with ED visits for syncope appears to have actually increased. There have been no apparent improvements in diagnostic yield for admissions. Novel strategies may be needed to change practice patterns for such patients.

Background Health-related social needs (HRSN) are increasingly recognized as important factors influencing healthcare outcomes and utilization. This study examined the association between unmet HRSNs and emergency department (ED) utilization among Medicare beneficiaries. Methods We conducted a retrospective, survey-weighted cohort analysis of the 2015–2016 Medicare Current Beneficiary Survey (MCBS) linked with Medicare fee-for-service claims. The study included beneficiaries aged ≥ 65 years enrolled in fee-for-service Medicare who completed the MCBS. The primary predictor was having ≥ 1 unmet HRSN (food insecurity, delaying care due to cost, or difficulty accessing medical care). Primary outcomes included an index ED visit (1 ED visit) and any ED revisit within one year (≥ 2 ED visits); hospital admission from these ED visits was a secondary outcome. We fit multivariable logistic regression models adjusted for demographic, socioeconomic, and clinical factors. Interactions were tested using adjusted Wald tests. Results Among 16,990 beneficiaries, 6.2% ( n  = 1,046) reported one or more unmet HRSNs. Within one year of completion of the survey, 27.7% ( n  = 4,702) had an 1 ED visit, with 9.03% of all beneficiaries ( n  = 1,535) requiring admission to the hospital. In adjusted analyses, beneficiaries with unmet HRSN had significantly higher odds of ≥ 2 ED visits (OR 1.47, 95% CI 1.12–1.91) compared to those without unmet HRSNs, but not for index ED visit. The oldest age category (85 + years) showed significantly increased odds of both index ED visits and revisits. Unmet HRSN were not significantly associated with risk of subsequent hospital admission for both index ED visit and ED revisit. Conclusion Self-reported unmet HRSNs were associated with significantly increased odds of ≥ 2 ED visits but not an 1 ED visit within one year of the MCBS survey. These findings highlight the importance of improved and standardized data collection of HRSNs to understand the impacts on ED utilization. Oldest age patients had increased odds of index ED visits and revisits. Further investigation should focus on strategies to reduce ED recidivism in vulnerable older populations.

Asthma is common, resulting in 53 million emergency department (ED) visits annually. Little is known about variation in cost and quality of ED asthma care. We sought to describe variation in costs and 7-day ED revisit rates for asthma care across EDs. Our primary objective was to test for an association between ED costs and the likelihood of a 7-day revisit for another asthma exacerbation. We used the 2014 Florida State Emergency Department Database to perform an observational study of ED visits by patients ≥18 years old with a primary diagnosis of asthma that were discharged home. We compared patient and hospital characteristics of index ED discharges with and without 7-day revisits, then tested the association between ED revisits and index ED costs. Multilevel regression was performed to account for hospital-level clustering. In 2014, there were 54,060 adult ED visits for asthma resulting in discharge, and 1667 (3%) were associated with an asthma-related ED revisit within 7 days. Median cost for an episode of ED asthma care was $597 with an interquartile range of $371–980. After adjusting for both patient and hospital characteristics, lack of insurance was associated with higher odds of revisit (OR 1.42, 95% CI 1.18–1.71), while private insurance, female gender, and older age were associated with lower odds of revisit. Hospital costs were not associated with ED revisits (OR = 1.00; 95% CI 1.00–1.00). Hospital costs associated with ED asthma visits vary but are not associated with odds of ED revisit.

Patients with labor and sex trafficking experiences seek healthcare while and after being trafficked. Their trafficking experiences are often unrecognized by clinicians who lack a validated tool to systematically screen for trafficking. We aimed to derive and validate a brief, comprehensive trafficking screening tool for use in healthcare settings. Patients were randomly selected to participate in this prospective study based on time of arrival. Data collectors administered 5 dichotomous index questions and a reference standard trafficking assessment tool that requires 30 to 60 minutes to administer. Data collection was from June 2016 to January 2021. Data from patients in 5 New York City (NYC) emergency departments (EDs) were used for tool psychometric derivation, and data from patients in a Fort Worth ED were used for external validation. Clinically stable ED adults (aged ≥18 years) were eligible to participate. Candidate questions were selected from the Trafficking Victim Identification Tool (TVIT). The study outcome measurement was a determination of a participant having a lifetime experience of labor and/or sex trafficking based on the interpretation of the reference standard interview, the TVIT. Overall, 4127 ED patients were enrolled. In the derivation group, the reference standard identified 36 (1.1%) as positive for a labor and/or sex trafficking experience. In the validation group, 12 (1.4%) were positive by the reference standard. Rapid Appraisal for Trafficking (RAFT) is a new 4‐item trafficking screening tool: in the derivation group, RAFT was 89% sensitive (95% confidence interval [CI], 79%–99%) and 74% specific (95% CI, 73%–76%) and in the external validation group, RAFT was 100% sensitive (95% CI, 100%–100%) and 61% specific (95% CI, 56%–65%). The rapid, 4‐item RAFT screening tool demonstrated good sensitivity compared with the existing, resource‐intensive reference standard tool. RAFT may enhance the detection of human trafficking in EDs. Additional multicenter studies and research on RAFT's implementation are needed.

Asthma affects 1 in every 12 persons in the United States, resulting in 1.9 million emergency department (ED) visits annually. However, the lack of patient-reported outcome measures (PROMs) validated for use in the ED limits the evaluation of interventions to improve ED asthma care. To address this knowledge gap, this study protocol will (1) develop and test the validity and reliability of the Patient Reported Outcomes for Acute Asthma Care and Treatment instrument (PROAACT), (2) test whether receiving more guideline-concordant ED care is associated with improved PROAACT responses, and (3) evaluate the association between PROAACT score and subsequent ED revisits and hospitalizations. This is a prospective cohort study of adult patients visiting the ED for acute asthma exacerbation across 3 EDs at an urban, tertiary care health system. Eligible patients are 18 years or older, have a prior diagnosis of asthma (self-reported or documented in the electronic health record), are English-speaking, and experiencing an ED visit for asthma exacerbation as determined by the treating clinician. Enrolled participants complete an initial PROM survey during their ED visit assessing their symptoms in the preceding 7 days, then complete a follow-up survey 7 days after ED discharge assessing changes in the symptoms in the subsequent 7 days. To test whether guideline-concordant care is associated with improved PROAACT scores, we will conduct a retrospective chart review of medications ordered during the ED visit, and then compare guideline adherence to changes in PROAACT scores. To test whether improved PROAACT scores are associated with fewer return ED visits and hospitalizations, we will extract all-cause ED visits and hospitalizations within 30 days from a regional health information exchange, and then compare usage to changes in PROAACT scores. We will use item response theory to develop scale responses based on summed item responses, which will allow us to test associations with clinical outcomes, including adherence to guideline-recommended care and return ED visits and hospitalizations. Recruitment is ongoing and has experienced numerous challenges related to the COVID-19 pandemic. To date, we have enrolled over 250 participants and have completed over 200 follow-ups. Recruitment is expected to conclude in spring 2025. Our study is intended to validate the use of PROMs during ED visits for acute asthma exacerbation among adult patients. Completion of the proposed aims will result in one of the first PROMs intended for use among adult ED patients and support the feasibility of collecting PROMs in the ED setting. Clinical Trials.gov NCT04349020; https://clinicaltrials.gov/study/NCT04349020. DERR1-10.2196/67195.

Background Sickle cell disease (SCD) is an autosomal recessive blood disorder affecting approximately 100,000 Americans and 3.1 million people globally. The scarcity of relevant knowledge and experience with rare diseases creates a unique need for cooperation and infrastructure to overcome challenges in translating basic research advances into clinical advances. Despite registry initiatives in SCD, the unavailability of descriptions of the selection process and copies of final data collection tools, coupled with incomplete representation of the SCD population hampers further research progress. This manuscript describes the SCDIC (Sickle Cell Disease Implementation Consortium) Registry development and makes the SCDIC Registry baseline and first follow-up data collection forms available for other SCD research efforts. Results Study data on 2400 enrolled patients across eight sites was stored and managed using Research Electronic Data Capture (REDCap). Standardized data collection instruments, recruitment and enrollment were refined through consensus of consortium sites. Data points included measures taken from a variety of validated sources (PHENX, PROMIS and others). Surveys were directly administered by research staff and longitudinal follow-up was coordinated through the DCC. Appended registry forms track medical records, event-related patient invalidation, pregnancy, lab reporting, cardiopulmonary and renal functions. Conclusions The SCDIC Registry strives to provide an accurate, updated characterization of the adult and adolescent SCD population as well as standardized, validated data collecting tools to guide evidence-based research and practice.

Social determinants of health (SDOH) are critical drivers of health disparities and patient outcomes. However, accessing and collecting patient-level SDOH data can be operationally challenging in the emergency department (ED) clinical setting, requiring innovative approaches. This scoping review examines the potential of AI and data science for modeling, extraction, and incorporation of SDOH data specifically within EDs, further identifying areas for advancement and investigation. We conducted a standardized search for studies published between 2015 and 2022, across Medline (Ovid), Embase (Ovid), CINAHL, Web of Science, and ERIC databases. We focused on identifying studies using AI or data science related to SDOH within emergency care contexts or conditions. Two specialized reviewers in emergency medicine (EM) and clinical informatics independently assessed each article, resolving discrepancies through iterative reviews and discussion. We then extracted data covering study details, methodologies, patient demographics, care settings, and principal outcomes. Of the 1047 studies screened, 26 met the inclusion criteria. Notably, 9 out of 26 (35%) studies were solely concentrated on ED patients. Conditions studied spanned broad EM complaints and included sepsis, acute myocardial infarction, and asthma. The majority of studies (n=16) explored multiple SDOH domains, with homelessness/housing insecurity and neighborhood/built environment predominating. Machine learning (ML) techniques were used in 23 of 26 studies, with natural language processing (NLP) being the most commonly used approach (n=11). Rule-based NLP (n=5), deep learning (n=2), and pattern matching (n=4) were the most commonly used NLP techniques. NLP models in the reviewed studies displayed significant predictive performance with outcomes, with F1-scores ranging between 0.40 and 0.75 and specificities nearing 95.9%. Although in its infancy, the convergence of AI and data science techniques, especially ML and NLP, with SDOH in EM offers transformative possibilities for better usage and integration of social data into clinical care and research. With a significant focus on the ED and notable NLP model performance, there is an imperative to standardize SDOH data collection, refine algorithms for diverse patient groups, and champion interdisciplinary synergies. These efforts aim to harness SDOH data optimally, enhancing patient care and mitigating health disparities. Our research underscores the vital need for continued investigation in this domain.

Despite increasing ED visits, evidence suggests overall hospitalization rates have decreased; however, it is unknown what clinical conditions account for these changes. We aim to describe condition-specific trends and hospital-level variation in hospitalization rates after ED visits from 2006 to 2014. Retrospective observational study of adult ED visits to U.S. acute care hospitals using nationally weighted data from the 2006–2014 National Emergency Department Survey. Our primary outcome was ED admission rate, defined as the number of admissions originating in the ED divided by the number of ED visits. We report admission rates overall and for each condition, including changes over time. We used logistic regression to compare the odds of ED admission from 2006 to 2014, adjusting for patient and hospital characteristics. We also measured hospital-level variation by calculating hospital-level median ED admission rates and interquartile ranges. After adjusting for patient and hospital characteristics, the odds of ED admission for any condition were 0.49 (CI 0.45, 0.52) in 2014 compared to 2006. The conditions with the greatest relative change in ED admission rates were chest pain (21.7 to 7.5%) and syncope (28.9 to 13.8%). The decline in ED admission rates were accompanied by increased variation in hospital-level ED admission rates. Recent reductions in ED admissions are largely attributable to decreased admissions for conditions amenable to outpatient critical pathways. Focusing on hospitals with persistently above-average ED admission rates may be a promising approach to improve the value of acute care.