Faculty Bibliography

IMPORTANCE/UNASSIGNED:Patient portal messaging volume has increased substantially and is linked to clinician burnout. In response, health systems are increasingly adopting large language models (LLMs) to generate draft replies to patient messages, yet little is known about how patients interpret and evaluate artificial intelligence (AI) involvement in this communication channel. OBJECTIVE/UNASSIGNED:To understand how patients perceive AI-drafted responses to electronic patient portal messages and to identify implications for patient-centered implementation. DESIGN, SETTING, AND PARTICIPANTS/UNASSIGNED:This qualitative study used vignette-based prompts mirroring scenarios from a prior survey on patient preferences regarding AI-drafted online messaging. Participants included a sample of 40 adult patients from a large academic health system who had previously been surveyed about AI-drafted portal messaging. Recruitment sought increased representation of individuals from minoritized racial and ethnic groups, older patients, and participants with lower satisfaction with AI-drafted messaging in prior survey responses. Data were collected between April and August 2025. EXPOSURE/UNASSIGNED:One-time interviews conducted via videoconference. Interviews included vignettes and draft messages varying in tone, length, and implied authorship, alongside questions about messaging preferences, AI, and health care. MAIN OUTCOMES AND MEASURES/UNASSIGNED:Patient perspectives regarding (1) portal messaging preferences, (2) comfort with AI-drafted replies, (3) preferred message tone and content, and (4) AI disclosure expectations. RESULTS/UNASSIGNED:This study of 40 patients included 30 women (75.0%), 9 men (22.5%), 14 Black or African American participants (35.0%), 13 White participants (32.5%), and 13 patients aged 65 years or older (32.5%). Patients described portal messaging as transactional, prioritizing timely problem resolution over relational depth. Patients expressed high comfort with AI-drafted messages, conditional on clinician oversight. Patient preferences for tone and expressions of empathy varied across individuals and contexts and were driven less by whether messages appeared AI-like or human-like than by whether tone, length, and detail fit the purpose and stakes of the communication. Participants broadly endorsed AI disclosure, describing transparency as important for trust, but differed in preferred disclosure timing and format. CONCLUSIONS AND RELEVANCE/UNASSIGNED:In this qualitative study of patient perspectives on AI-drafted portal messaging, patients supported AI use when it improved communication efficiency, so long as clinicians reviewed and remained accountable for messages. Implementation should prioritize clinician oversight, context-sensitive communication standards, standardized disclosure practices, and monitoring for downstream effects on quality, equity, and patient trust and safety.

PURPOSE/UNASSIGNED:Underutilization of genetic testing among Hispanic males results in higher rates of variants of uncertain significance (VUS). We examined the impact of VUS on decision making and behavioral intentions. METHODS/UNASSIGNED:We conducted a nationwide survey of 807 US Hispanic males aged ≥40 in English and Spanish on perspectives about genetic testing results. Logistic regression was used to examine predictors of worry and behavior change with a hypothetical VUS result. RESULTS/UNASSIGNED:Over half of Hispanic male participants would still participate in genetic testing with a 1 in 5 chance of VUS. However, 36% were at least somewhat likely to regret testing and 49.9% would worry about cancer risk with VUS results. In addition, 56.3% were somewhat or very likely to change behavior due to a VUS, such as getting checked by the doctor more often or telling family members to get checked. Younger age and college education were associated with more worry and intended behavior change. CONCLUSION/UNASSIGNED:Although many Hispanic males are interested in genetic testing despite the higher likelihood of VUS, potential consequences include decisional regret, anxiety, and even changes in behavior. Effective counseling and support are important for minoritized groups undergoing genetic evaluation to avoid the potential to exacerbate health disparities.

IMPORTANCE:Artificial intelligence (AI)-assisted ambient documentation technologies that use audio from clinic visits to generate documentation are being deployed across hospital systems to optimize clinical note writing and reduce administrative burden. However, little is known about optimal approaches to engage patients in informed consent for these tools. OBJECTIVES:To explore clinician and patient experiences with consent processes and examine the association between the use of ambient documentation tools and the patient-clinician relationship. DESIGN, SETTING, AND PARTICIPANTS:This study was conducted from March 1 to December 31, 2024, in ambulatory practices across specialties in a large urban academic health center as part of an ongoing operational quality improvement initiative. Participants included clinicians and patients participating in an operational proof-of-concept exploration of ambient documentation technology. MAIN OUTCOMES AND MEASURES:A pragmatic, sequential, inductive-deductive qualitative evaluation was conducted of informed consent contexts, processes, and challenges among ambulatory clinicians and patients exposed to ambient documentation technology. Evaluation included site visits, clinical observations, clinician interviews, and patient surveys to explore perceptions and challenges related to consent. RESULTS:A total of 121 ambient documentation pilot users included 18 clinicians (mean [SD] years of practice, 18.6 [100]; 10 men [55.6%]) and 103 patients (mean [SD] age, 37 [12.5] years; 65 women [63.1%]). The most common consent approach was a verbal patient-clinician conversation prior to an individual encounter. Patients and clinicians had a spectrum of comfort with ambient technology; 77 patients (74.8%) reported being comfortable or very comfortable with their physician using ambient documentation. Patient trust, detail in the consent discussion, and intended tool use were associated with patient comfort and intent to consent. Technical understanding was associated with comfort with consent conversations: when provided basic information about the technology, 84 patients (81.6%) consented; this decreased to 57 patients (55.3%) when details about AI features, data storage, and corporate involvement were disclosed. Perceived benefits included reduced documentation burden, improved decision-making, and enhanced communication. Concerns included data security, legal liability, cognitive impacts, and equity. When asked about responsibility for medical errors linked to ambient documentation, 66 patients (64.1%) held physicians accountable; for data security breaches, 79 patients (76.7%) believed vendors should be responsible. Participants suggested a flexible consent model with digital touchpoints, education, nonclinical staff support, and opt-out options. CONCLUSIONS AND RELEVANCE:In this quality improvement study of 121 pilot users of AI-assisted ambient documentation technology, informed consent relied primarily on verbal conversations that varied based on time, knowledge, and the patient-clinician relationship. A flexible, multimodal approach-including education, time to discuss risks and benefits, digital resources, nonclinical staff involvement, and clear opt-out options-may improve consent processes and support broader acceptance of ambient documentation tools.

INTRODUCTION:As the utility of genomic sequencing increases, its use in healthcare will continue to expand beyond expert clinics toward nonspecialist practices such as primary care. At the same time, discordance in genetic variant identification and classification between laboratories remains a concern for the field. This research assesses how clinicians with and without genetics expertise understand and trust genetic test results, underscoring how variation in the handling of genetic test results can have real impact on patient care. METHODS:We conducted 40 interviews with genetics experts, including clinical geneticists and genetic counselors, and nonexpert clinicians including primary care providers and cardiologists. RESULTS:Clinical geneticists and genetic counselors reported spending significant time assessing the validity of results from genetic testing laboratories, conversing with laboratories about those results, and potentially reinterpreting results. Conversely, primary care providers and cardiologists without specific genetics expertise reported high levels of trust in laboratory accuracy and variant interpretation, and did not reassess results. CONCLUSION:We find significant variation in how genetics experts and nonexperts understand the trustworthiness of genetic laboratory reports. This variation could lead to differences in patient care between clinical settings and requires additional guidance for clinicians regarding the handling of genetic test results.

Research institutions often lack policies addressing the risks and benefits of enrolling "invested parties" such as investigators, research staff, and patient, caregiver, and community representatives (groups most affected by a disease or intervention) in studies where they have direct involvement. Invested parties may have both strong motivations to study the condition or intervention and to participate as study subjects. More guidance is needed to promote appropriate access to research participation and mitigate potential risks. This article addresses the gap in guidance by presenting an ethical framework and practical guidelines for the enrollment of invested parties. Drawing from experiences with the Researching COVID to Enhance Recovery (RECOVER) Initiative, a large multisite observational cohort study, we argue that invested parties should not be categorically excluded from enrollment in their own research studies if certain criteria are met and appropriate safeguards are in place. We underscore the need to balance inclusion with fairness, promote valid voluntary informed consent, ensure data privacy, protect scientific validity, and mitigate unique risks to invested parties as participants. Additionally, we recommend regular reporting and empirical assessment to evaluate the impact of enrolling invested parties on participants and study outcomes.