Why do users override alerts? Utilizing large language model to summarize comments and optimize clinical decision support.

OBJECTIVES: To evaluate the capability of using generative artificial intelligence (AI) in summarizing alert comments and to determine if the AI-generated summary could be used to improve clinical decision support (CDS) alerts. MATERIALS AND METHODS: We extracted user comments to alerts generated from September 1, 2022 to September 1, 2023 at Vanderbilt University Medical Center. For a subset of 8 alerts, comment summaries were generated independently by 2 physicians and then separately by GPT-4. We surveyed 5 CDS experts to rate the human-generated and AI-generated summaries on a scale from 1 (strongly disagree) to 5 (strongly agree) for the 4 metrics: clarity, completeness, accuracy, and usefulness. RESULTS: Five CDS experts participated in the survey. A total of 16 human-generated summaries and 8 AI-generated summaries were assessed. Among the top 8 rated summaries, five were generated by GPT-4. AI-generated summaries demonstrated high levels of clarity, accuracy, and usefulness, similar to the human-generated summaries. Moreover, AI-generated summaries exhibited significantly higher completeness and usefulness compared to the human-generated summaries (AI: 3.4 ± 1.2, human: 2.7 ± 1.2, P = .001). CONCLUSION: End-user comments provide clinicians' immediate feedback to CDS alerts and can serve as a direct and valuable data resource for improving CDS delivery. Traditionally, these comments may not be considered in the CDS review process due to their unstructured nature, large volume, and the presence of redundant or irrelevant content. Our study demonstrates that GPT-4 is capable of distilling these comments into summaries characterized by high clarity, accuracy, and completeness. AI-generated summaries are equivalent and potentially better than human-generated summaries. These AI-generated summaries could provide CDS experts with a novel means of reviewing user comments to rapidly optimize CDS alerts both online and offline.

VEXAS-Defining UBA1 Somatic Variants in 245,368 Diverse Individuals in the NIH All Of Us Cohort.

OBJECTIVE: Somatic variants in UBA1 cause VEXAS, a recently described, systemic autoinflammatory disease. Research on VEXAS has largely focused on highly symptomatic patients. We sought to determine the prevalence of canonical, VEXAS-associated somatic variants and their disease penetrance in a diverse, unselected population. METHODS: We analyzed clinical-grade whole genome sequencing data from 245,368 participants in the All of Us Research Program. We compared persons carrying a canonical VEXAS-associated somatic variant to age, sex, and ancestry matched controls across the domains of diagnoses, medications, and laboratory values. RESULTS: 74 participants were identified who carry one VEXAS-defining somatic variant, UBA1 c.121A>C, p.Met41Leu. The variant allele fraction ranged from 4.5% to 33%. No other canonical VEXAS-associated variants were identified. Of the 74 carriers, 62 (84%) were women, 20 (27%) were African American, and 14 (19%) were American Admixed/Latino. There was no statistically significant association between case/control status and any VEXAS-associated diagnosis code, medication prescription, or laboratory value. CONCLUSION: We report the largest cohort to date of persons with the VEXAS-associated p.Met41Leu somatic variant. This cohort differed substantially from reported cohorts of patients with clinical VEXAS, having a higher proportion of persons who were young, female, and of diverse ancestry. Variant allele fractions were lower than reported in clinical VEXAS cohorts, and bioinformatic analysis detected no clinical manifestations of VEXAS. Thus, the UBA1 p.Met41Leu somatic variant displayed incomplete penetrance for VEXAS. Further study is needed to determine the natural history of VEXAS-associated somatic variants in the predisease phase.

Triglyceride Concentrations and Their Relationship to Sedation Choice and Outcomes in Mechanically Ventilated Patients Receiving Propofol.

Rationale: Propofol is a first-line sedative agent in the intensive care unit (ICU) but may be associated with hypertriglyceridemia and pancreatitis. To date, the relationship between propofol-induced hypertriglyceridemia and pancreatitis, as well as clinician responses to propofol-induced hypertriglyceridemia, have not been comprehensively studied. Objectives: To assess the incidence of hypertriglyceridemia and pancreatitis in patients receiving continuous propofol infusions in the ICU and to describe the association between hypertriglyceridemia and the use of nonpropofol continuous sedative infusions. Methods: This was a retrospective observational cohort study conducted at three urban academic hospitals within a single health system. Findings were additionally validated using the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database containing data from a separate tertiary care hospital. Mechanically ventilated adult patients who received a continuous propofol infusion between 2016 and 2021 were included. The primary exposure was serum triglyceride concentration, and hypertriglyceridemia was defined as a triglyceride concentration greater than 400 mg/dl. Outcomes included new-onset pancreatitis as well as receipt of midazolam, dexmedetomidine, or ketamine after the triglyceride measurement. The incidence of pancreatitis was compared between groups using a Fisher's Exact test. Multivariable logistic regression was used to assess the association between dichotomized triglyceride concentration and alternative sedative use. Results: In the primary cohort of 7,037 patients, 1,724 (24.5%) had one or more triglyceride concentration measured. Of these, 1,365 (79.2%) had a maximum concentration of less than 400 mg/dl, and 359 (20.8%) had a maximum concentration of greater than 400 mg/dl. Compared with patients with low triglyceride concentrations, patients with high triglyceride concentrations were more likely to receive a continuous infusion of midazolam (37.0% vs. 16.4%; adjusted odds ratio [aOR], 3.1; 95% confidence interval [CI], 2.2-4.4; P < 0.01), ketamine (22.8% vs. 6.9%; aOR, 3.5; 95% CI, 2.3-5.3; P < 0.01), and dexmedetomidine (57.7% vs. 46.6%; aOR, 1.5; 95% CI, 1.1-2.0; P < 0.01). Rates of midazolam infusion increased as triglyceride concentrations exceeded 500 mg/dl. Forty-four (0.6%) patients developed pancreatitis after propofol initiation, of which 4 (9.1%) were considered related to propofol-associated hypertriglyceridemia. Findings were similar in the MIMIC-IV cohort. Conclusions: Propofol-associated hypertriglyceridemia is relatively common in mechanically ventilated ICU patients who have triglycerides measured. Pancreatitis related to propofol-associated hypertriglyceridemia is rare. Patients who develop hypertriglyceridemia while receiving propofol are more likely to receive continuous infusions of other sedatives.

Epidemiology of Organ Failure Before and During COVID-19 Pandemic Surge Conditions.

BACKGROUND: Understanding the distribution of organ failure before and during the COVID-19 pandemic surge can provide a deeper understanding of how the pandemic strained health care systems and affected outcomes. OBJECTIVE: To assess the distribution of organ failure in 3 New York City hospitals during the COVID-19 pandemic. METHODS: A retrospective cohort study of adult admissions across hospitals from February 1, 2020, through May 31, 2020, was conducted. The cohort was stratified into those admitted before March 17, 2020 (prepandemic) and those admitted on or after that date (SARS-CoV-2-positive and non-SARS-CoV-2). Sequential Organ Failure Assessment scores were computed every 2 hours for each admission. RESULTS: A total of 1 794 975 scores were computed for 20 704 admissions. Before and during the pandemic, renal failure was the most common type of organ failure at admission and respiratory failure was the most common type of hospital-onset organ failure. The SARS-CoV-2-positive group showed a 231% increase in respiratory failure compared with the prepandemic group. More than 65% of hospital-onset organ failure in the prepandemic group and 83% of hospital-onset respiratory failure in the SARS-CoV-2-positive group occurred outside intensive care units. The SARS-CoV-2-positive group showed a 341% increase in multiorgan failure compared with the prepandemic group. Compared with the prepandemic and non-SARS-CoV-2 patients, SARS-CoV-2-positive patients had significantly higher mortality for the same admission and maximum organ failure score. CONCLUSION: Most hospital-onset organ failure began outside intensive care units, with a marked increase in multiorgan failure during pandemic surge conditions and greater hospital mortality for the severity of organ failure.

Trial by fire: How physicians responding to the COVID-19 pandemic illuminated the need for digital credentials.

The current credentialing process for physicians struggled to accommodate fluctuating regional demands for providers during the severe acute respiratory syndrome coronavirus 2 pandemic. This hurdle highlighted existing inefficiencies and difficulties facing healthcare systems across the world and led us to explore how credentialing can be improved using digital technologies. We explain how this is a critical moment to make the shift from physical to digital credentials by specifying how a digital credentialing system could simplify onboarding for providers, enable secure expansion of telehealth services, and enhance information exchange.

Authenticating Health Activity Data Using Distributed Ledger Technologies.

The on-demand digital healthcare ecosystem is on the near horizon. It has the potential to extract a wealth of information from "big data" collected at the population level, to enhance preventive and precision medicine at the patient level. This may improve efficiency and quality while decreasing cost of healthcare delivered by professionals. However, there are still security and privacy issues that need to be addressed before algorithms, data, and models can be mobilized safely at scale. In this paper we discuss how distributed ledger technologies can play a key role in advancing electronic health, by ensuring authenticity and integrity of data generated by wearable and embedded devices. We demonstrate how the Masked Authenticated Messaging extension module of the IOTA protocol can be used to securely share, store, and retrieve encrypted activity data using a tamper-proof distributed ledger.

A retrospective quasi-experimental study of a transitional housing program for patients with severe and persistent mental illness.

Transitional housing programs aim to improve living skills and housing stability for tenuously housed patients with mental illness. 113 consecutive Transitional Housing Team (THT) patients were matched to 139 controls on diagnosis, time of presentation, gender and prior psychiatric hospitalisation and compared using a difference-in-difference analysis for illness acuity and service use outcomes measured 1 year before and after THT entry/exit. There was a statistically significant difference-in-difference favouring THT participants for bed days (mean difference in difference -20.76 days, SE 9.59, p = 0.031) and living conditions (HoNOS Q11 mean difference in difference -0.93, SE 0.23, p < 0.001). THT cost less per participant (I$14,024) than the bed-days averted (I$17,348). The findings of reductions in bed days and improved living conditions suggest that transitional housing programs can have a significant positive impact for tenuously housed patients with high inpatient service usage, as well as saving costs for mental health services.

A retrospective quasi-experimental study of a community crisis house for patients with severe and persistent mental illness.

OBJECTIVE: There is increasing international evidence that crisis houses can reduce the time spent in acute psychiatric inpatient units for patients with severe and persistent mental illness, at a lower cost and in an environment preferable to patients. We evaluated the Alternatives to Hospitalisation (AtH) program, a crisis house operating in outer suburban Brisbane. METHODS: One hundred and ninety-three AtH patients were compared to 371 matched controls admitted to a peer hospital district acute psychiatric unit. Hospitalisations, demographics and illness acuity were compared one year before and after an acute index episode of residential care involving hospital and/or AtH. Hospital bed-days during the index episode were compared between AtH participants and controls. The cost of bed-days averted was compared to the cost of providing the AtH program. RESULTS: AtH participants spent 5.35 fewer days in hospital during the index episode than controls, after adjustment for illness acuity, living conditions, marital status and emergency department (ED) presentations. Per patient cost of averted psychiatric inpatient bed-days, $5948.22, was higher than the per patient cost of providing AtH, $3071.44. AtH participants had higher levels of illness acuity, ED presentations and acute psychiatric admissions than controls in the year after the index episode. CONCLUSIONS: For acutely unwell, stably housed patients, able to be managed outside of a secure facility, a crisis house program can reduce acute psychiatric bed-days, providing a cost saving for mental health services.