Using generative AI to investigate medical imagery models and datasets.

BACKGROUND: AI models have shown promise in performing many medical imaging tasks. However, our ability to explain what signals these models have learned is severely lacking. Explanations are needed in order to increase the trust of doctors in AI-based models, especially in domains where AI prediction capabilities surpass those of humans. Moreover, such explanations could enable novel scientific discovery by uncovering signals in the data that aren't yet known to experts. METHODS: In this paper, we present a workflow for generating hypotheses to understand which visual signals in images are correlated with a classification model's predictions for a given task. This approach leverages an automatic visual explanation algorithm followed by interdisciplinary expert review. We propose the following 4 steps: (i) Train a classifier to perform a given task to assess whether the imagery indeed contains signals relevant to the task; (ii) Train a StyleGAN-based image generator with an architecture that enables guidance by the classifier ("StylEx"); (iii) Automatically detect, extract, and visualize the top visual attributes that the classifier is sensitive towards. For visualization, we independently modify each of these attributes to generate counterfactual visualizations for a set of images (i.e., what the image would look like with the attribute increased or decreased); (iv) Formulate hypotheses for the underlying mechanisms, to stimulate future research. Specifically, present the discovered attributes and corresponding counterfactual visualizations to an interdisciplinary panel of experts so that hypotheses can account for social and structural determinants of health (e.g., whether the attributes correspond to known patho-physiological or socio-cultural phenomena, or could be novel discoveries). FINDINGS: To demonstrate the broad applicability of our approach, we present results on eight prediction tasks across three medical imaging modalities-retinal fundus photographs, external eye photographs, and chest radiographs. We showcase examples where many of the automatically-learned attributes clearly capture clinically known features (e.g., types of cataract, enlarged heart), and demonstrate automatically-learned confounders that arise from factors beyond physiological mechanisms (e.g., chest X-ray underexposure is correlated with the classifier predicting abnormality, and eye makeup is correlated with the classifier predicting low hemoglobin levels). We further show that our method reveals a number of physiologically plausible, previously-unknown attributes based on the literature (e.g., differences in the fundus associated with self-reported sex, which were previously unknown). INTERPRETATION: Our approach enables hypotheses generation via attribute visualizations and has the potential to enable researchers to better understand, improve their assessment, and extract new knowledge from AI-based models, as well as debug and design better datasets. Though not designed to infer causality, importantly, we highlight that attributes generated by our framework can capture phenomena beyond physiology or pathophysiology, reflecting the real world nature of healthcare delivery and socio-cultural factors, and hence interdisciplinary perspectives are critical in these investigations. Finally, we will release code to help researchers train their own StylEx models and analyze their predictive tasks of interest, and use the methodology presented in this paper for responsible interpretation of the revealed attributes. FUNDING: Google.

Care Partner Engagement in Secure Messaging Between Patients With Diabetes and Their Clinicians: Cohort Study.

BACKGROUND: Patient engagement with secure messaging (SM) via digital patient portals has been associated with improved diabetes outcomes, including increased patient satisfaction and better glycemic control. Yet, disparities in SM uptake exist among older patients and racial and ethnic underserved groups. Care partners (family members or friends) may provide a means for mitigating these disparities; however, it remains unclear whether and to what extent care partners might enhance SM use. OBJECTIVE: We aim to examine whether SM use differs among older patients with diabetes based on the involvement of care partner proxies. METHODS: This is a substudy of the ECLIPPSE (Employing Computational Linguistics to Improve Patient-Provider Secure Emails) project, a cohort study taking place in a large, fully integrated health care delivery system with an established digital patient portal serving over 4 million patients. Participants included patients with type 2 diabetes aged ≥50 years, newly registered on the patient portal, who sent ≥1 English-language message to their clinician between July 1, 2006, and December 31, 2015. Proxy SM was identified by having a registered proxy. To identify nonregistered proxies, a computational linguistics algorithm was applied to detect words and phrases more likely to appear in proxy messages compared to patient-authored messages. The primary outcome was the annual volume of secure messages (sent or received); secondary outcomes were the length of time to the first SM sent by patient or proxy and the number of annual SM exchanges (unique message topics generating ≥1 reply). RESULTS: The mean age of the cohort (N=7659) at this study's start was 61 (SD 7.16) years; 75% (n=5573) were married, 15% (n=1089) identified as Black, 10% (n=747) Chinese, 12% (n=905) Filipino, 13% (n=999) Latino, and 30% (n=2225) White. Further, 49% (n=3782) of patients used a proxy to some extent. Compared to nonproxy users, proxy users were older (P<.001), had lower educational attainment (P<.001), and had more comorbidities (P<.001). Adjusting for patient sociodemographic and clinical characteristics, proxy users had greater annual SM volume (20.7, 95% CI 20.2-21.2 vs 10.9, 95% CI 10.7-11.2; P<.001), shorter time to SM initiation (hazard ratio vs nonusers: 1.30, 95% CI 1.24-1.37; P<.001), and more annual SM exchanges (6.0, 95% CI 5.8-6.1 vs 2.9, 95% CI 2.9-3.0, P<.001). Differences in SM engagement by proxy status were similar across patient levels of education, and racial and ethnic groups. CONCLUSIONS: Among a cohort of older patients with diabetes, proxy SM involvement was independently associated with earlier initiation and increased intensity of messaging, although it did not appear to mitigate existing disparities in SM. These findings suggest care partners can enhance patient-clinician telecommunication in diabetes care. Future studies should examine the effect of care partners' SM involvement on diabetes-related quality of care and clinical outcomes.

Patient Acceptability of Telehealth Medication Abortion Care in the United States, 2021‒2022: A Cohort Study.

Objectives. Despite the recent expansion of direct-to-patient telehealth abortion care in the United States, patient experiences with the service are not well understood. Methods. We described care experiences of 1600 telehealth abortion patients in 2021 to 2022 and used logistic regression to explore differences by race or ethnicity and between synchronous (phone or video) and asynchronous (secure messaging) telehealth abortion care. Results. Most patients trusted the provider (98%), felt telehealth was the right decision (96%), felt cared for (92%), and were very satisfied (89%). Patients most commonly cited privacy (76%), timeliness (74%), and staying at home (71%) as benefits. The most commonly reported drawback was initial uncertainty about whether the service was legitimate (38%). Asian patients were less likely to be very satisfied than White patients (79% vs 90%; P = .008). Acceptability was high for both synchronous and asynchronous care. Conclusions. Telehealth abortion care is highly acceptable, and benefits include privacy and expediency. Public Health Implications. Telehealth abortion can expand abortion access in an increasingly restricted landscape while maintaining patient-centered care. (Am J Public Health. 2024;114(2):241-250. https://doi.org/10.2105/AJPH.2023.307437).

A Closed-Loop Falls Monitoring and Prevention App for Multiple Sclerosis Clinical Practice: Human-Centered Design of the Multiple Sclerosis Falls InsightTrack.

BACKGROUND: Falls are common in people with multiple sclerosis (MS), causing injuries, fear of falling, and loss of independence. Although targeted interventions (physical therapy) can help, patients underreport and clinicians undertreat this issue. Patient-generated data, combined with clinical data, can support the prediction of falls and lead to timely intervention (including referral to specialized physical therapy). To be actionable, such data must be efficiently delivered to clinicians, with care customized to the patient's specific context. OBJECTIVE: This study aims to describe the iterative process of the design and development of Multiple Sclerosis Falls InsightTrack (MS-FIT), identifying the clinical and technological features of this closed-loop app designed to support streamlined falls reporting, timely falls evaluation, and comprehensive and sustained falls prevention efforts. METHODS: Stakeholders were engaged in a double diamond process of human-centered design to ensure that technological features aligned with users' needs. Patient and clinician interviews were designed to elicit insight around ability blockers and boosters using the capability, opportunity, motivation, and behavior (COM-B) framework to facilitate subsequent mapping to the Behavior Change Wheel. To support generalizability, patients and experts from other clinical conditions associated with falls (geriatrics, orthopedics, and Parkinson disease) were also engaged. Designs were iterated based on each round of feedback, and final mock-ups were tested during routine clinical visits. RESULTS: A sample of 30 patients and 14 clinicians provided at least 1 round of feedback. To support falls reporting, patients favored a simple biweekly survey built using REDCap (Research Electronic Data Capture; Vanderbilt University) to support bring-your-own-device accessibility-with optional additional context (the severity and location of falls). To support the evaluation and prevention of falls, clinicians favored a clinical dashboard featuring several key visualization widgets: a longitudinal falls display coded by the time of data capture, severity, and context; a comprehensive, multidisciplinary, and evidence-based checklist of actions intended to evaluate and prevent falls; and MS resources local to a patient's community. In-basket messaging alerts clinicians of severe falls. The tool scored highly for usability, likability, usefulness, and perceived effectiveness (based on the Health IT Usability Evaluation Model scoring). CONCLUSIONS: To our knowledge, this is the first falls app designed using human-centered design to prioritize behavior change and, while being accessible at home for patients, to deliver actionable data to clinicians at the point of care. MS-FIT streamlines data delivery to clinicians via an electronic health record-embedded window, aligning with the 5 rights approach. Leveraging MS-FIT for data processing and algorithms minimizes clinician load while boosting care quality. Our innovation seamlessly integrates real-world patient-generated data as well as clinical and community-level factors, empowering self-care and addressing the impact of falls in people with MS. Preliminary findings indicate wider relevance, extending to other neurological conditions associated with falls and their consequences.

Email-Based Recruitment Into the Health eHeart Study: Cohort Analysis of Invited Eligible Patients.

BACKGROUND: Web- or app-based digital health studies allow for more efficient collection of health data for research. However, remote recruitment into digital health studies can enroll nonrepresentative study samples, hindering the robustness and generalizability of findings. Through the comprehensive evaluation of an email-based campaign on recruitment into the Health eHeart Study, we aim to uncover key sociodemographic and clinical factors that contribute to enrollment. OBJECTIVE: This study sought to understand the factors related to participation, specifically regarding enrollment, in the Health eHeart Study as a result of a large-scale remote email recruitment campaign. METHODS: We conducted a cohort analysis on all invited University of California, San Francisco (UCSF) patients to identify sociodemographic and clinical predictors of enrollment into the Health eHeart Study. The primary outcome was enrollment, defined by account registration and consent into the Health eHeart Study. The email recruitment campaign was carried out from August 2015 to February 2016, with electronic health record data extracted between September 2019 and December 2019. RESULTS: The email recruitment campaign delivered at least 1 email invitation to 93.5% (193,606/206,983) of all invited patients and yielded a 3.6% (7012/193,606) registration rate among contacted patients and an 84.1% (5899/7012) consent rate among registered patients. Adjusted multivariate logistic regression models analyzed independent sociodemographic and clinical predictors of (1) registration among contacted participants and (2) consent among registered participants. Odds of registration were higher among patients who are older, women, non-Hispanic White, active patients with commercial insurance or Medicare, with a higher comorbidity burden, with congestive heart failure, and randomized to receive up to 2 recruitment emails. The odds of registration were lower among those with medical conditions such as dementia, chronic pulmonary disease, moderate or severe liver disease, paraplegia or hemiplegia, renal disease, or cancer. Odds of subsequent consent after initial registration were different, with an inverse trend of being lower among patients who are older and women. The odds of consent were also lower among those with peripheral vascular disease. However, the odds of consent remained higher among patients who were non-Hispanic White and those with commercial insurance. CONCLUSIONS: This study provides important insights into the potential returns on participant enrollment when digital health study teams invest resources in using email for recruitment. The findings show that participant enrollment was driven more strongly by sociodemographic factors than clinical factors. Overall, email is an extremely efficient means of recruiting participants from a large list into the Health eHeart Study. Despite some improvements in representation, the formulation of truly diverse studies will require additional resources and strategies to overcome persistent participation barriers.

Visit and Between-Visit Interaction Frequency Before and After COVID-19 Telehealth Implementation.

Importance: Telehealth implementation associated with the COVID-19 public health emergency (PHE) affected patient-clinical team interactions in numerous ways. Yet, studies have narrowly examined billed patient-clinician visits rather than including visits with other team members (eg, pharmacists) or between-visit interactions. Objective: To evaluate rates of change over time in visits (in-person, telehealth) and between-visit interactions (telephone calls, patient portal messages) overall and by key patient characteristics. Design, Setting, and Participants: This retrospective cohort study included adults with diabetes receiving primary care at urban academic (University of California San Francisco [UCSF]) and safety-net (San Francisco Health Network [SFHN]) health care systems. Encounters from April 2019 to March 2021 were analyzed. Exposure: Telehealth implementation over 3 periods: pre-PHE (April 2019 to March 2020), strict shelter-in-place (April to June 2020), and hybrid-PHE (July 2020 to March 2021). Main Outcomes and Measures: The main outcomes were rates of change in monthly mean number of total encounters, visits with any health care team member, visits with billing clinicians, and between-visit interactions. Key patient-level characteristics were age, race and ethnicity, language, and neighborhood socioeconomic status (nSES). Results: Of 15 148 patients (4976 UCSF; 8975 SFHN) included, 2464 (16%) were 75 years or older, 7734 (51%) were female patients, 9823 (65%) self-identified as racially or ethnically minoritized, 6223 (41%) had a non-English language preference, and 4618 (31%) lived in the lowest nSES quintile. After accounting for changes to care delivery through an interrupted time-series analysis, total encounters increased in the hybrid-PHE period (UCSF: 2.3% per patient/mo; 95% CI, 1.6%-2.9% per patient/mo; SFHN: 1.8% per patient/mo, 95% CI, 1.3%-2.2% per patient/mo), associated primarily with growth in between-visit interactions (UCSF: 3.1% per patient/mo, 95% CI, 2.3%-3.8% per patient/mo; SFHN: 2.9% per patient/mo, 95% CI, 2.3%-3.4% per patient/mo). In contrast, rates of visits were stable during the hybrid-PHE period. Although there were fewer differences in visit use by key patient-level characteristics during the hybrid-PHE period, pre-PHE differences in between-visit interactions persisted during the hybrid-PHE period at SFHN. Asian and Chinese-speaking patients at SFHN had fewer monthly mean between-visit interactions compared with White patients (0.46 [95% CI, 0.42-0.50] vs 0.59 [95% CI, 0.53-0.66] between-visit interactions/patient/mo; P < .001) and English-speaking patients (0.52 [95% CI, 0.47-0.58] vs 0.61 [95% CI, 0.56-0.66] between-visit interactions/patient/mo; P = .03). Conclusions and Relevance: In this study, pre-PHE growth in overall patient-clinician encounters persisted after PHE-related telehealth implementation, driven in both periods by between-visit interactions. Differential utilization based on patient characteristics was observed, which may indicate disparities. The implications for health care team workload and patient outcomes are unknown, particularly regarding between-visit interactions. Therefore, to comprehensively understand care utilization for patients with chronic diseases, research should expand beyond billed visits.

Social Determinants of Health Documentation in Structured and Unstructured Clinical Data of Patients With Diabetes: Comparative Analysis.

Background: Electronic health records (EHRs) have yet to fully capture social determinants of health (SDOH) due to challenges such as nonexistent or inconsistent data capture tools across clinics, lack of time, and the burden of extra steps for the clinician. However, patient clinical notes (unstructured data) may be a better source of patient-related SDOH information. Objective: It is unclear how accurately EHR data reflect patients' lived experience of SDOH. The manual process of retrieving SDOH information from clinical notes is time-consuming and not feasible. We leveraged two high-throughput tools to identify SDOH mappings to structured and unstructured patient data: PatientExploreR and Electronic Medical Record Search Engine (EMERSE). Methods: We included adult patients (≥18 years of age) receiving primary care for their diabetes at the University of California, San Francisco (UCSF), from January 1, 2018, to December 31, 2019. We used expert raters to develop a corpus using SDOH in the compendium as a knowledge base as targets for the natural language processing (NLP) text string mapping to find string stems, roots, and syntactic similarities in the clinical notes of patients with diabetes. We applied advanced built-in EMERSE NLP query parsers implemented with JavaCC. Results: We included 4283 adult patients receiving primary care for diabetes at UCSF. Our study revealed that SDOH may be more significant in the lives of patients with diabetes than is evident from structured data recorded on EHRs. With the application of EMERSE NLP rules, we uncovered additional information from patient clinical notes on problems related to social connectionsisolation, employment, financial insecurity, housing insecurity, food insecurity, education, and stress. Conclusions: We discovered more patient information related to SDOH in unstructured data than in structured data. The application of this technique and further investment in similar user-friendly tools and infrastructure to extract SDOH information from unstructured data may help to identify the range of social conditions that influence patients' disease experiences and inform clinical decision-making.

Serious Mental Illness, Glycemic Control, and Neighborhood Factors within an Urban Diabetes Cohort.

BACKGROUND AND HYPOTHESIS: Serious mental illness (SMI) may compromise diabetes self-management. This study assessed the association between SMI and glycemic control, and explored sociodemographic predictors and geographic clustering of this outcome among patients with and without SMI. STUDY DESIGN: We used electronic health record data for adult primary care patients with diabetes from 2 San Francisco health care delivery systems. The primary outcome was poor glycemic control (hemoglobin A1c >9.0%), which was modeled on SMI diagnosis status and sociodemographics. Geospatial analyses examined hotspots of poor glycemic control and neighborhood characteristics. STUDY RESULTS: The study included 11 694 participants with diabetes, 21% with comorbid SMI, of whom 22% had a schizophrenia spectrum or bipolar disorder. Median age was 62 years; 52% were female and 79% were Asian, Black, or Hispanic. In adjusted models, having schizophrenia spectrum disorder or bipolar disorder was associated with greater risk for poor glycemic control (vs participants without SMI, adjusted relative risk [aRR] = 1.24; 95% confidence interval, 1.02, 1.49), but having broadly defined SMI was not. People with and without SMI had similar sociodemographic correlates of poor glycemic control including younger versus older age, Hispanic versus non-Hispanic White race/ethnicity, and English versus Chinese language preference. Hotspots for poor glycemic control were found in neighborhoods with more lower-income, Hispanic, and Black residents. CONCLUSIONS: Poor diabetes control was significantly related to having a schizophrenia spectrum or bipolar disorder, and to sociodemographic factors and neighborhood. Community-based mental health clinics in hotspots could be targets for implementation of diabetes management services.

"At the End I Have a Say": Engaging the Chinese Community in Advance Care Planning.

CONTEXT: Despite the association of advance care planning (ACP) with improved patient and caregiver outcomes, Chinese American elders have low rates of ACP. OBJECTIVES: Assess ACP facilitators/barriers in the San Francisco (SF) Chinese community and codesign, implement, and test community-based ACP-promoting pilot events. METHODS: A Chinese Community Committee (N = 19 community-based organization leaders, health system representatives, community members) conducted focus groups in Cantonese and English with Chinese older adults (age ≥55), caregivers, and community leaders. The Committee designed and implemented pilot events in-person and online. We analyzed focus group data using thematic analysis; assessed pre-to-post-event readiness to engage in ACP (validated survey; 14 scale, 4 = most ready); and assessed event acceptability. RESULTS: A total of 34 people participated in six focus groups. Themes described Chinese community-specific importance of ACP (e.g., reduces family burden), barriers (e.g., younger generations lack tools to discuss ACP with elders and vice versa), and facilitators (e.g., intergenerational events, culturally/linguistically appropriate materials). Based on focus groups findings, the Committee developed a novel ACP tool and designed intergenerational events. A total of 195 participants attended 10 events; 95% were Chinese, 90% spoke Chinese languages, 80% were women. ACP readiness increased significantly (1.66 [SD 0.84] vs. 2.03 [SD 0.85]; P < 0.001); 94% of participants were comfortable attending and 96% would recommend events. CONCLUSION: Community-developed intergenerational events that highlight the value of ACP and address barriers are acceptable and increase ACP engagement in the Chinese community.

Language-Specific Challenges and Solutions for Equitable Telemedicine Implementation in the Primary Care Safety Net During COVID-19.

PURPOSE: Telemedicine care dramatically expanded during the COVID-19 pandemic. We characterized facilitators and barriers to telemedicine implementation among safety-net primary care clinics serving patients with limited English proficiency (LEP). METHODS: We collected data on telemedicine volume and patient demographics among safety-net clinics participating in a telemedicine learning collaborative. Data on various metrics were reported to the collaborative from February 2019 through August 2021. We conducted semi-structured interviews with clinical and quality leaders, purposively sampling clinics serving high proportions of patients with LEP. We analyzed interviews with a mixed inductive-deductive approach applying the Consolidated Framework for Implementation Research. RESULTS: By September 2020, the 23 sites served 121,589 unique patients with in-person and 120,338 with telephone visits; 47% of these patients had LEP. Of 10,897 unique patients served by video visits, 38% had LEP. As a proportion of total visits, telemedicine (telephone and video) visits increased from 0-17% in October 2019-March 2020 to 10-98% in March-August 2020. We conducted 14 interviews at 11 sites. Themes included (1) existing telemedicine platforms and interpreter services were not optimized to support patients with LEP; (2) clinics invested significant labor iterating workflows; (3) sites with technological infrastructure and language-concordant staff were best suited to serve patients; (4) patients speaking less-represented languages or experiencing intersecting literacy barriers were underserved with telemedicine. Interviewees recommended innovations in telemedicine platforms and community-based access. CONCLUSIONS: Safety-net sites relied on existing resources to accommodate patients with LEP, but struggled providing access for the most marginalized. Proactive, data-driven strategies to address patient and community barriers as well as optimize clinical workflows with high-quality, certified medical interpreters are needed to ensure equitable access.

Usability and acceptability of virtual reality for chronic pain management among diverse patients in a safety-net setting: a qualitative analysis.

Objective: The aim of this study was to understand the usability and acceptability of virtual reality (VR) among a racially and ethnically diverse group of patients who experience chronic pain. Materials and Methods: Using the Technology Acceptance Model theory, we conducted semistructured interviews and direct observation of VR use with English-speaking patients who experience chronic pain treated in a public healthcare system (n = 15), using a commercially available VR technology platform. Interviews included questions about current pain management strategies, technology use, experiences and opinions with VR, and motivators for future use. Results: Before the study, none of the 15 participants had heard about or used VR for pain management. Common motivators for VR use included a previous history of substance use and having exhausted many other options to manage their pain and curiosity. Most participants had a positive experience with VR and 47% found that the VR modules distracted them from their pain. When attempting the navigation-based usability tasks, most participants (73%-92%) were able to complete them independently. Discussion: VR is a usable tool for diverse patients with chronic pain. Our findings suggest that the usability of VR is not a barrier and perhaps a focus on improving the accessibility of VR in safety-net settings is needed to reduce disparities in health technology use. Conclusions: The usability and acceptability of VR are rarely studied in diverse patient populations. We found that participants had a positive experience using VR, showed interest in future use, and would recommend VR to family and friends.

Health system approaches and experiences implementing the 4Ms: Insights from 3 early adopter health systems.

BACKGROUND: Redesigning the healthcare system to consistently provide effective and tailored care to older adults is needed. The 4Ms (What Matters, Mobility, Medication, and Mentation) offer a framework to guide health systems' efforts to deliver Age-Friendly care. We use an implementation science framework to characterize and assess real-world implementation experiences with the 4Ms across varied health systems. METHODS: With expert input, we selected three health systems that were early adopters of the 4Ms and engaged in different implementation support models through the Institute for Healthcare Improvement. We conducted 29 semi-structured interviews with diverse stakeholders from each site. Stakeholders ranged from hospital leadership to frontline clinicians. Interviews covered each site's approach to and experiences with implementation, including facilitators and barriers. Interviews were recorded, transcribed, and deductively coded using the Consolidated Framework for Implementation Research. We characterized each site's implementation decisions and then inductively identified overarching themes and subthemes with supporting quotes. RESULTS: Health systems varied in their implementation approach, including the implementation order of each of the 4Ms. We identified three overarching themes: (1) the 4Ms offered a compelling conceptual framework for advancing Age-Friendly care, but implementation was complex and fragmented; (2) complete and sustained implementation of the 4Ms required multidisciplinary and multilevel leadership and engagement; (3) strategies that facilitate implementation success and support frontline culture change included top-down communication and infrastructure alongside hands-on clinical education and support. Common barriers are siloed implementation efforts across settings that impeded synergies and scaling; disengaged physicians; and difficulty implementing What Matters in a meaningful way. CONCLUSIONS: Similar to other implementation studies, we identified multifactorial domains impacting 4Ms implementation. To achieve Age-Friendly transformation, health systems must plan for and attend to multiple phases of implementation while ensuring that the work coheres under a unified vision that spans disciplines and settings.

Acceptability and Usability of a Wearable Device for Sleep Health Among English- and Spanish-Speaking Patients in a Safety Net Clinic: Qualitative Analysis.

BACKGROUND: Sleep disorders are common and disproportionately affect marginalized populations. Technology, such as wearable devices, holds the potential to improve sleep quality and reduce sleep disparities, but most devices have not been designed or tested with racially, ethnically, and socioeconomically diverse patients. Inclusion and engagement of diverse patients throughout digital health development and implementation are critical to achieving health equity. OBJECTIVE: This study aims to evaluate the usability and acceptability of a wearable sleep monitoring device-SomnoRing-and its accompanying mobile app among patients treated in a safety net clinic. METHODS: The study team recruited English- and Spanish-speaking patients from a mid-sized pulmonary and sleep medicine practice serving publicly insured patients. Eligibility criteria included initial evaluation of obstructed sleep apnea, which is most appropriate for limited cardiopulmonary testing. Patients with primary insomnia or other suspected sleep disorders were not included. Patients tested the SomnoRing over a 7-night period and participated in a 1-hour semistructured web-based qualitative interview covering perceptions of the device, motivators and barriers to use, and general experiences with digital health tools. The study team used inductive or deductive processes to code interview transcripts, guided by the Technology Acceptance Model. RESULTS: A total of 21 individuals participated in the study. All participants owned a smartphone, almost all (19/21) felt comfortable using their phone, and few already owned a wearable (6/21). Almost all participants wore the SomnoRing for 7 nights and found it comfortable. The following four themes emerged from qualitative data: (1) the SomnoRing was easy to use compared to other wearable devices or traditional home sleep testing alternatives, such as the standard polysomnogram technology for sleep studies; (2) the patient's context and environment, such as family and peer influence, housing status, access to insurance, and device cost affected the overall acceptance of the SomnoRing; (3) clinical champions motivated use in supporting effective onboarding, interpretation of data, and, ongoing technical support; and (4) participants desired more assistance and information to best interpret their own sleep data summarized in the companion app. CONCLUSIONS: Racially, ethnically, and socioeconomically diverse patients with sleep disorders perceived a wearable as useful and acceptable for sleep health. Participants also uncovered external barriers related to the perceived usefulness of the technology, such as housing status, insurance coverage, and clinical support. Future studies should further examine how to best address these barriers so that wearables, such as the SomnoRing, can be successfully implemented in the safety net health setting.

Racial and sociodemographic predictors of COVID-19 compared with influenza, appendicitis, and all-cause hospitalization: retrospective cohort analysis.

OBJECTIVE: To determine whether inequities in COVID-19 infection and hospitalization differ from those for common medical conditions: influenza, appendicitis, and all-cause hospitalization. DESIGN: Retrospective study based on electronic health records of three healthcare systems in San Francisco (university, public, and community) examining (1) racial/ethnic distribution in cases and hospitalization among patients with diagnosed COVID-19 (March-August 2020) and patients with diagnosed influenza, diagnosed appendicitis, or all-cause hospitalization (August 2017-March 2020), and (2) sociodemographic predictors of hospitalization among those with diagnosed COVID-19 and influenza. RESULTS: Patients 18 years or older with diagnosed COVID-19 (N = 3934), diagnosed influenza (N = 5932), diagnosed appendicitis (N = 1235), or all-cause hospitalization (N = 62,707) were included in the study. The age-adjusted racial/ethnic distribution of patients with diagnosed COVID-19 differed from that of patients with diagnosed influenza or appendicitis for all healthcare systems, as did hospitalization from these conditions compared to any cause. For example, in the public healthcare system, 68% of patients with diagnosed COVID-19 were Latine, compared with 43% of patients with diagnosed influenza, and 48% of patients with diagnosed appendicitis (p < 0.05). In multivariable logistic regressions, COVID-19 hospitalizations were associated with male sex, Asian and Pacific Islander race/ethnicity, Spanish language, and public insurance in the university healthcare system, and Latine race/ethnicity and obesity in the community healthcare system. Influenza hospitalizations were associated with Asian and Pacific Islander and other race/ethnicity in the university healthcare system, obesity in the community healthcare system, and Chinese language and public insurance in both the university and community healthcare systems. CONCLUSIONS: Racial/ethnic and sociodemographic inequities in diagnosed COVID-19 and hospitalization differed from those for diagnosed influenza and other medical conditions, with consistently higher odds among Latine and Spanish-speaking patients. This work highlights the need for disease-specific public health efforts in at-risk communities in addition to structural upstream interventions.

"We've Got to Bring Information to Where People Are Comfortable": Community-Based Advance Care Planning with the Black Community.

BACKGROUND: People identifying as Black/African American are less likely to engage in advance care planning (ACP) compared to their White peers, despite the association of ACP with improved patient and caregiver outcomes. OBJECTIVES: Assess facilitators/barriers to ACP in the San Francisco (SF) Black community and co-design/implement/test community-based ACP pilot events. DESIGN: Community-based participatory research, including qualitative research, intervention development, and implementation. PARTICIPANTS: In partnership with the SF Palliative Care Workgroup (which includes health system, city, and community-based organizations), we formed an African American Advisory Committee (n = 13). We conducted 6 focus groups with Black older adults (age ≥ 55), caregivers, and community leaders (n = 29). The Advisory Committee then selected 5 community-based organizations through a widespread request for proposal. These community-based organizations designed and implemented community-based pilot events to support ACP engagement. MAIN MEASURES: Two authors analyzed recorded focus group transcripts using thematic analysis. We assessed pre- vs post-event readiness to engage in ACP (validated ACP Engagement Survey; 1-4 scale, 4 = most ready) using Wilcoxon signed rank tests and assessed event acceptability with open-ended questions. KEY RESULTS: Themes included the importance of ACP to the Black community (sub-themes: strengthens families; preserves dignity, particularly for sexual/gender minorities; is tied to financial planning) and facilitators for increasing ACP engagement (sub-themes: culturally relevant materials; events in trusted community spaces including Black-owned businesses). A total of 114 participants attended 5 events; 74% identified as Black, and 16% as sexual/gender minorities. Readiness to engage in ACP was similar pre- vs post-events; 98% would recommend the events to others. CONCLUSIONS: Community-based ACP events designed and led by and for the Black community are highly acceptable. Novel insights underscored the importance of financial planning as part of ACP and the role of Black-owned businesses as trusted spaces for ACP-related discussions.

Neighborhood-level COVID vaccination and booster disparities: A population-level analysis across California.

Objectives: To describe vaccine and booster uptake by neighborhood-level factors in California. Methods: We examined trends in COVID-19 vaccination up to September 21, 2021, and boosters up to March 29, 2022 using data from the California Department of Public Health. Quasi-Poisson regression was used to model the association between neighborhood-level factors and fully vaccinated and boosted among ZIP codes. Sub-analyses on booster rates were compared among the 10 census regions. Results: In a minimally adjusted model, a higher proportion of Black residents was associated with lower vaccination (HR = 0.97; 95%CI: 0.96-0.98). However, in a fully adjusted model, proportion of Black, Hispanic/Latinx, and Asian residents were associated with higher vaccination rates (HR = 1.02; 95%CI: 1.01-1.03 for all). The strongest predictor of low vaccine coverage was disability (HR = 0.89; 95%CI: 0.86-0.91). Similar trends persisted for booster doses. Factors associated with booster coverage varied by region. Conclusions: Examining neighborhood-level factors associated with COVID-19 vaccination and booster rates uncovered significant variation within the large and geographically and demographically diverse state of California. Equity-based approaches to vaccination must ensure a robust consideration of multiple social determinants of health.

Randomized trial protocol for remote monitoring for equity in advancing the control of hypertension in safety net systems (REACH-SNS) study.

BACKGROUND: Self-measured blood pressure monitoring (SMBP) is essential to effective management of hypertension. This study aims to evaluate effectiveness and implementation of SMBP that leverages: cellular-enabled home BP monitors without a need for Wi-Fi or Bluetooth; simple communication modalities such as text messaging to support patient engagement; and integration into existing team-based workflows in safety-net clinics. METHODS: This study will be conducted with patients in San Francisco who are treated within a network of safety-net clinics. English and Spanish-speaking patients with diagnosed hypertension will be eligible for the trial if they have recent BP readings ≥140/90 mmHg and do not have co-morbid conditions that make home BP monitoring more complex to manage. This study will implement a three-arm randomized controlled trial to compare varying levels of implementation support: 1) cellular-enabled BP monitors (with minimal implementation support), 2) cellular-enabled BP monitors with protocol-based implementation support (text reminders for patients; aggregated BP summaries sent to primary care providers), and 3) cellular-enabled BP monitors and pharmacist-led support (pharmacist coaching and independent medication adjustments). RESULTS: For the main analysis, we will use mixed effects linear regression to compare the change in primary outcome of systolic BP. Secondary outcomes include BP control (<140/90 mmHg), medication intensification, patient-reported outcomes, and implementation processes (i.e., engagement with the intervention). DISCUSSION: This study will design and test a digital health intervention for use in marginalized populations treated within safety net settings, evaluating both effectiveness and implementation to advance more equitable health outcomes.

Ratings and experiences in using a mobile application to increase physical activity among university students: implications for future design.

University students have low levels of physical activity and are at risk of mental health disorders. Mobile apps to encourage physical activity can help students, who are frequent smartphone-users, to improve their physical and mental health. Here we report students' qualitative feedback on a physical activity smartphone app with motivational text messaging. We provide recommendations for the design of future apps. 103 students used the app for 6 weeks in the context of a clinical trial (NCT04440553) and answered open-ended questions before the start of the study and at follow-up. A subsample (n = 39) provided additional feedback via text message, and a phone interview (n = 8). Questions focused on the perceived encouragement and support by the app, text messaging content, and recommendations for future applications. We analyzed all transcripts for emerging themes using qualitative coding in Dedoose. The majority of participants were female (69.9%), Asian or Pacific Islander (53.4%), with a mean age of 20.2 years, and 63% had elevated depressive symptoms. 26% felt encouraged or neutral toward the app motivating them to be more physically active. Participants liked messages on physical activity benefits on (mental) health, encouraging them to complete their goal, and feedback on their activity. Participants disliked messages that did not match their motivations for physical activity and their daily context (e.g., time, weekday, stress). Physical activity apps for students should be adapted to their motivations, changing daily context, and mental health issues. Feedback from this sample suggests a key to effectiveness is finding effective ways to personalize digital interventions.