Publications

BACKGROUND: Emergency department (ED) boarding exposes admitted older adults to a high-disruption environment. Sleep loss during acute hospitalization is associated with adverse outcomes and may contribute to delirium, yet objective sleep measurement in this setting is difficult. Consumer multisensor wearables may permit low-burden, longitudinal measurement, but the feasibility and acceptability of ring-based monitoring among acutely ill older adults is unknown.

OBJECTIVE: This study aimed to estimate the feasibility and acceptability of Oura Ring Gen 3 sleep monitoring in older adults admitted to the hospital from the ED, and to describe exploratory device-measured and patient-reported sleep outcomes.

METHODS: We conducted a single-site prospective cohort study from February through March 2025. Patients aged 65 years or older with an inpatient medical bed request were recruited in the ED and asked to wear the ring continuously for up to 4 days or until discharge; monitoring continued after transfer to an inpatient unit from the ED. Participants underwent daily 3-Minute Diagnostic Interview for Confusion Assessment Method-Defined Delirium (3D-CAM) assessments, cognitive testing when not delirious, daily sleep surveys, and an exit survey. Feasibility outcomes were the approached-to-enrolled proportion and the proportion of study completers with wear for more than 75% of the analyzed period. Acceptability was the proportion of exit respondents who disagreed or strongly disagreed that the ring bothered them. Device-estimated sleep duration was normalized to hours per 24 hours, and its association with mean patient-reported sleep quality was assessed using Kendall τ.

RESULTS: Of 29 patients approached for consent, 15 enrolled (51.7%, 95% CI 32.5%-70.6%); 1 withdrew immediately, leaving 14 who completed study procedures. Of these 14 patients, 13 wore the ring for ≥75% of their enrollment period (92.9%, 95% CI 66.1%-99.8%). Eleven participants completed the exit survey, and 9 reported that the ring was not bothersome (81.8%, 95% CI 48.2%-97.7%). Among the 13 participants who wore the ring for ≥75% of the analyzed period, median device-estimated sleep was 4.3 (IQR 2.6-7.0) hours per 24 hours. Participants reported good or very good sleep on 9 of 19 surveyed nights (47.4%). Mean sleep quality and device-estimated sleep duration were positively associated among 10 participants with complete paired data (Kendall τ=0.54; P=.04).

CONCLUSIONS: In this pilot study, we found that sleep monitoring with the Oura Ring Gen 3 was feasible and acceptable to older adults admitted from the ED. These findings support larger studies validating the Oura Ring against polysomnography in hospitalized older adults.

BACKGROUND: Telehealth use surged during the COVID-19 pandemic, but utilization after the pandemic and whether differences in use by various sociodemographic and clinical factors have persisted are unknown.

OBJECTIVE: To assess changes in telehealth use among US adults from 2022 to 2024.

DESIGN AND PARTICIPANTS: This repeated cross-sectional study used nationally representative data from the 2022 and 2024 Health Information National Trends Survey (HINTS).

MAIN MEASURES: Self-reported telehealth use (any video or telephone visit in the past 12 months). Changes in the use were evaluated using logistic regression, adjusting for sociodemographic, clinical, and access-related variables and interactions with survey year.

KEY RESULTS: Among 11,386 respondents (mean [SD] age, 48.6 [17.8] years; 51.1% female), the unadjusted prevalence of telehealth use declined from 39.0% (95% CI [36.9%, 41.3%]) in 2022 to 34.9% (95% CI [32.1%, 37.9%]) in 2024, with a substantial decline among women. The proportion of visits conducted by video remained stable (71.1%, 95% CI [68.8%, 73.4%]), but increased considerably among uninsured individuals. Overall telehealth use was associated with age, internet use, higher income, and large-metro residence; among users, video use was additionally associated with Northeast residence and depression. Most associations did not differ by year, while gender differences in overall use and insurance differences in video use narrowed. Although approximately 20% reported technical problems, over 75% rated telehealth care as comparable to in-person care.

CONCLUSIONS: Self-reported telehealth use declined modestly after 2022, although video-based visits and perceived quality remained stable. Most of the previously recognized differences in telehealth use by sociodemographic and clinical factors persisted over time, but differences by gender and insurance status narrowed. Due to high-perceived quality, broad access to telehealth should continue with efforts to reduce structural and technological barriers.

BACKGROUND: General medicine (GM) research fellowships facilitate health care professionals to obtain the skills and mentored research experience needed for a research career. Yet, little is known about the experience and outcomes of recent graduates of GM fellowships.

OBJECTIVE: To survey (1) GM research fellows and recent graduates and (2) program directors to learn about the opportunities and challenges facing GM research fellowships.

DESIGN: Cross-sectional retrospective survey study.

PARTICIPANTS: Individuals who were enrolled in or completed a GM research fellowship (identified through prior fellow lists, advertising from the Society of General Internal Medicine, and fellowship program directors) between 2012 and 2022 in which ≥ 50% of effort was devoted to training and conducting research; 2022 program directors were recruited separately.

MAIN MEASURES: Respondents were asked to rate factors motivating them to pursue a research fellowship and their experiences during fellowship and to report their professional outcomes. Program directors were asked to rate how challenging different fellowship activities were to them.

KEY RESULTS: A total of 160 (42.2%) current/former fellows and 27 (67.5%) program directors completed surveys. Fellowship training was considered essential to obtain the skills necessary to become a GM researcher. Fellows were highly productive (median of 4 papers [IQR 3-6] published from fellowship work); the majority obtained research faculty positions. Gaps in training included negotiation skills, team management, and grant writing. Trainees and graduates sought more cross-institutional collaboration during fellowship. Approximately 1/3 reported at least one symptom of burnout. Fellowship outcomes generally did not vary by gender or under-represented in medicine status. Program directors (88%) found fellowship recruitment challenging.

CONCLUSIONS: Research fellowships play a critical role in cultivating GM investigators. Consistent investment in GM research fellowships is needed to ensure a pipeline of researchers with the skills necessary to improve general medicine, especially during an ongoing crisis in primary care.

Risk prediction model development has expanded rapidly, but few models have been translated into patient-facing tools that support informed decision-making. Existing breast cancer models provide no guidance on how to incorporate risk into decisions around screening or prevention medications, limiting their practical utility. To address this gap, we previously developed and validated a competing risk regression model that simultaneously predicts breast cancer and non-breast cancer death and then developed a web-based decision aid application that integrates this model with interactive, personalized information on screening and prevention medications. Using this application as a case study, we present a framework for developing an online decision aid using R Shiny. While prior Shiny tutorials have focused on simple applications or calculators, practical guidance for integrating risk prediction into multi-page, interactive decision support tools remains limited. In our tutorial, we describe key components of development, including application structure, user input collection, real-time calculation of individualized risk estimates, and presentation of results in a clear, interpretable format. We also demonstrate implementation of core Shiny functionalities, including reactive values for dynamic updates, data visualization techniques to contextualize risk estimates, and use of the observeEvent function to enable conditional display and navigation. Through this tutorial, we illustrate how risk calculators can be extended into comprehensive, dynamic and clinically useful tools that support informed decision-making.

STUDY OBJECTIVES: To compare end-of-life predictions as measured by the physician-answered surprise question (SQ), "Would you be surprised if this patient died in the next 6 months?"), the Geriatric End-of-Life Screening Tool (GEST) artificial intelligence (AI) model, and a new collaborative GEST+SQ model for predicting 6-month mortality in older emergency department (ED) patients.

METHODS: This was a single-site prospective cohort study (Nov 2022 to June 2023) at a tertiary academic ED of patients aged 65 years and older. Answers to the SQ were collected within the electronic health record at ED disposition and GEST scores were calculated from available records using laboratory, vital signs, demographic and historical data. Six-month mortality was adjudicated via electronic health record and state records. SQ and GEST were compared using sensitivity and specificity. A new logistic regression model was developed combining SQ and GEST (GEST+SQ) and compared with GEST alone, using area under receiver-operating characteristic curves (ROC-AUC) for discrimination and expected calibration error for calibration. We modeled a sequential screening pathway where low- and high-risk patients received only GEST screening, whereas intermediate-risk patients received both GEST and SQ, reporting the proportion of patients for whom adding the SQ to GEST would change a theoretical referral to intervention.

RESULTS: From 9,256 eligible patients, 3,479 had SQ responses (37.6%), with 13.3% 6-month mortality. When matching GEST sensitivity to SQ (83.8%), GEST had greater specificity than the SQ (61.5% [56.7 to 67.1] vs. 50.8% [49.1 to 52.6]). At matching specificity (50.8%), GEST sensitivity (90.0% [87.0 to 92.7]) exceeded the SQ (83.8% [80.3 to 87.0]). GEST had an receiver-operating characteristic - area under the curve (ROC-AUC) of 0.79 (0.77 to 0.81), whereas the GEST+SQ model had ROC-AUC of 0.80 (0.78 to 0.82). The GEST+SQ model had significantly improved expected calibration error of 0.01 (0.01 to 0.02) for GEST+SQ vs. 0.042 (0.03 to 0.05) for GEST alone. In a sequential screening pathway, as few as 5% of patients required SQ screening following GEST risk scoring.

CONCLUSION: GEST modestly outperformed the SQ for predicting 6-month mortality. A GEST+SQ collaborative model did not improve discrimination (ROC-AUC) over GEST alone, but improved calibration. Sequential screening using GEST and then the SQ for intermediate-risk patients could decrease physician screening burden by 95% relative to manual, SQ-only screening. Collaborative approaches integrating automated tools with targeted physician input may enhance ED mortality risk assessment while reducing clinician effort.

BACKGROUND: Whether to continue breast cancer screening beyond age 74 is uncertain. Decision aids may improve understanding of health information and support informed screening decisions. The goal of this study was to develop a video-based decision aid for breast cancer screening among older women using patient-centered design.

METHODS: Following the Framework for Innovation, the research team first used formative focus groups to understand older women's perspectives on mammography. We developed a prototype video based on decision aid best practices and formative focus group findings. We then evaluated the content, clarity, and style of the decision aid in cognitive testing focus groups. We made iterative changes to the video in response to focus group feedback. Focus groups included women age ≥ 70 without a personal history of breast cancer from Connecticut-area community and clinical settings. We coded and analyzed transcripts using both abductive and deductive approaches.

RESULTS: We convened 6 formative focus groups and 7 cognitive testing groups with 31 participants (mean age 78 [range 70-93]); 39% Black, 58% White, and 3% Latina. In focus groups, participants perceived screening as largely beneficial and saw overdiagnosis as unfamiliar. Some participants valued quantitative information about risks and benefits of screening, while others relied on experience, perceptions of risk, and beliefs about the efficacy of mammography to make screening decisions. We incorporated these perspectives into the framing, language, and narrative arc of the decision aid. In cognitive testing focus groups, participants found the decision aid informative and engaging.

DISCUSSION: Using a patient-centered approach, we developed a video-based decision aid for breast cancer screening for older women. Our design, which drew on the perspectives of older women, was perceived as easy to understand and informative. We will assess the impact of the decision aid on decision quality, decisional conflict, and intention to screen in future work.