The Uneven Distribution of Patient Portal Messages Across Patients and Physicians

Presentation Type: Oral Presentation - Regular
Presentation Time: 08:00 AM - 08:12 AM

Abstract Keywords: Health Policy, Reimbursement and Affordability, and Sustainability, Clinician Well-Being, Workforce Automation, Communication, and Workflow Efficiency
Primary Track: Advancing Wellness for Providers and Community with Consideration of Human Factors


Patient-initiated medical advice request (PMAR) messages are a growing source of uncompensated work for physicians, but how this work is distributed across patients and clinicians is poorly understood. Using patient-level data from a large academic health system and physician-level metadata from a national EHR vendor, we characterized the distribution of portal message volume across specialties and ambulatory visit volume. PMAR volume was highly skewed: a small fraction of patients and physicians accounted for a large share of messages. These results highlight the need for targeted informatics interventions and payment models that recognize the uneven burden of asynchronous portal-based care.

Speaker(s):
A J Holmgren, PhD
University of California, San Francisco

Author(s):
A J Holmgren, PhD - University of California, San Francisco; Nate Apathy, PhD - University of Maryland; Ateev Mehrotra, MD MPH - Brown University; Michael Chernew, PhD - Harvard University;

---

Characterizing Variation in Patient Provider Messaging Workflows Across Patient Groups

Presentation Type: Oral Presentation - Regular
Presentation Time: 08:12 AM - 08:24 AM

Abstract Keywords: Telemedicine, Health at Home, and Virtual Care, Workforce Automation, Communication, and Workflow Efficiency, Outcomes Improvement and Equity
Primary Track: Advancing Wellness for Providers and Community with Consideration of Human Factors


With an increased use of secure electronic messaging between patients and their primary care teams following the COVID pandemic, it is critical to understand the patterns of interaction downstream of an initial message and how these patterns vary by patient characteristics to assess and improve health system equity. Using UCSF’s Epic Clarity database, we derived data regarding messages sent by patients to their attending PCP in 2021 and the replies from their primary care teams. Across a sample of 38,060 patients who exchanged one of 495,531 messages with one of 184 primary care physicians, collectively exchanging 495,531 messages, we found significant differences in the prevalence of responses to patient messages by patient race/ethnicity and insurance. Black patients and those with Medicaid insurance or dual Medicare-Medicaid eligibility were less likely to receive a response to their message in 3 business days, but patients with Medicaid or dual Medicare-Medicaid insurance were more likely to have a visit after sending a message. These findings demonstrate how EHR messaging data can grant an understanding of electronic messaging interaction patterns and can help inform how to optimally meet the digital engagement needs of diverse populations.

Speaker(s):
Lisa Rotenstein, MD, MBA, MSc
UCSF

Author(s):
Lisa Rotenstein, MD, MBA, MSc - UCSF; Melanie Molina, MD, MAS - University of California, San Francisco; Sienna Kurland, MD - UCSF; Mitchell Tang, PhD - Columbia University, Mailman School of Public Health; Christopher Toretsky, MPH - University of California, San Francisco; Jade Christey, BA - UCSF; Robert Thombley, MS - UCSF; Michael Barnett, MD - Harvard University;

---

Conquer the In-Basket - one folder at a time

Presentation Type: Oral Presentation - Regular
Presentation Time: 08:24 AM - 08:36 AM

Abstract Keywords: Clinician Well-Being, Workforce Automation, Communication, and Workflow Efficiency, Telemedicine, Health at Home, and Virtual Care, Leadership and Strategy, Generative AI in Clinical Workflow: Ambient Listening, Chart Summarization, Automated Response with LLM, Outcomes Improvement and Equity
Primary Track: Advancing Wellness for Providers and Community with Consideration of Human Factors


Physicians and advanced practice providers in ambulatory practice spend a significant amount of time performing asynchronous tasks outside of patient-facing clinical hours. These tasks are collectively known as the "Inbox" or "In Basket". Ever growing concerns regarding provider work burden and the subsequent development of burnout have brought increased scrutiny on workflows inside the electronic health record (EHR). We describe Intermountain Health's journey to implementing a strategy comprised of an Inboxology team.

Speaker(s):
Jay Shi, MD
Intermountain Health

Author(s):
Jay Shi, MD - Intermountain Health; Jen House, MBA - Intermountain Health;

---

Addressing Electronic Inbox Burden in Primary Care: A Multisite Descriptive Study

Presentation Type: Oral Presentation - Regular
Presentation Time: 08:36 AM - 08:48 AM

Abstract Keywords: Clinician Well-Being, Generative AI in Clinical Workflow: Ambient Listening, Chart Summarization, Automated Response with LLM, Analytical Artificial Intelligence: ML, Digital Pathology, Imaging AI, Predictive Analytics, Governance
Primary Track: Advancing Wellness for Providers and Community with Consideration of Human Factors


Primary care physicians (PCPs) face increasing electronic inbox burdens, largely driven by patient messaging through electronic health records (EHRs), which contributes to PCP burnout and challenges the sustainability and quality of primary care in the United States. This multisite, mixed-method descriptive study describes team-based and artificial intelligence (AI) approaches to alleviating inbox workload across five large academic primary care networks. The study surveyed and interviewed primary care and clinical informatics leaders to characterize electronic inbox pooling workflows, team member roles, and the adoption of AI-drafted responses. Among 71 clinics representing over 550 PCPs, the most prevalent workflows involved messages being routed to medical assistants or administrative pools before reaching physicians, with detailed algorithms guiding non-physician staff in message handling. Supplemental supports included pharmacy technician pools and advanced practice provider teams for time-sensitive requests. Workflow evaluation methods varied, ranging from qualitative feedback to quantitative metrics such as message volume and refill rates. All networks trialed AI-drafted inbox responses, but uptake was low (8.6%–10.9%), with PCPs citing issues of voice, accuracy, and verbosity. In conclusion, while team-based pooling workflows are common, they have proven to be insufficient in addressing inbox burden and are often supplemented by centralized support. AI-drafted responses have not substantially reduced PCP workload. More rigorous evaluations may help inform which of these workflows can best fit the needs of individual health systems seeking to alleviate their PCPs’ electronic inbox burden.

Speaker(s):
Lisa Rotenstein, MD, MBA, MSc
UCSF

Author(s):
Lisa Rotenstein, MD, MBA, MSc - UCSF; Sienna Kurland, MD - University of California San Francisco; A J Holmgren, PhD - University of California, San Francisco; Alexandra King, MD - Cedars-Sinai Hospital System; Christine Iannaccone, MPH - Brigham and Women's Hospital; Michelle Frits, BA - Brigham and Women's Hospital; Lalima Hoq, MD, MPH - Cedars Sinai; Jacqueline You, MD, MBI - Mass General Brigham; Angela Rui, MA - Mass General Brigham; Edward Melnick, MD - Yale University, School of Medicine; Brian Williams, MD - Northeast Medical Group; Laurel Buchanan, MPH - Yale School of Medicine; Joshua Pevnick, MD, MSHS - Cedars-Sinai Medical Center; Teryl Nuckols, MD - Cedars Sinai Medical Center; Jennifer Haas, MD - Massachusetts General Hospital; Moshe Rancier, MD - Mass General Brigham; David Bates, MD - Mass General Brigham; Harvard University;

---

Implementation of an Electronic Closed-Loop Communication System for Critical Lab Results

Presentation Type: Oral Presentation - Regular
Presentation Time: 08:48 AM - 09:00 AM

Abstract Keywords: Workforce Automation, Communication, and Workflow Efficiency, Clinical Decision Support and Care Pathways, Human Factors and Usability, Clinician Well-Being, Health Policy, Reimbursement and Affordability, and Sustainability
Primary Track: Advancing Wellness for Providers and Community with Consideration of Human Factors


We deployed an electronic closed-loop system for critical lab result notifications, replacing manual phone calls with automated alerts to ordering providers. Post-implementation, over 80% of alerts were acknowledged within 10 minutes, with an average response under two minutes, saving over 400 staff hours monthly. This low-cost, scalable solution improved turnaround times, streamlined workflows, and enhanced patient safety, demonstrating effective integration of EHR notification technology across a large pediatric healthcare enterprise.

Speaker(s):
Eric Shelov, MD MBI
The Children's Hospital of Philadelphia/ Perelman School of Medicine, Univ. of Pennsylvania

Author(s):
Eric Shelov, MD MBI - The Children's Hospital of Philadelphia/ Perelman School of Medicine, Univ. of Pennsylvania; Amrom Obstfeld, MD PhD - Childrens Hospital of Philadelphia; Tracey Polsky, MD PhD - Children's Hospital of Philadelphia;

---

Retrieval-Augmented Guardrails for Error Detection in LLM-Generated Patient Portal Messages

Presentation Type: Oral Presentation - Regular
Presentation Time: 09:00 AM - 09:12 AM

Abstract Keywords: Generative AI in Clinical Workflow: Ambient Listening, Chart Summarization, Automated Response with LLM, Clinical Decision Support and Care Pathways, Data Privacy, Cybersecurity, Reliability, and Security, Quality Informatics and Lean, Outcomes Improvement and Equity
Primary Track: Big Data for Health


Rising volumes of patient portal messages have accelerated adoption of LLM-generated draft replies, yet clinicians must still review drafts for accuracy. We introduce a 5-domain, 59-code clinical error ontology and a Retrieval-Augmented Error Checking (RAEC) pipeline that uses similar historical messages to improve error detection, achieving higher concordance (50% vs 33%) and F1 (0.500 vs 0.256) compared with baseline, enabling automated guardrails that are better aligned with local clinical workflows and practice patterns.

Speaker(s):
Wenyuan Chen, MS
Stanford

Author(s):
Wenyuan Chen, MS - Stanford; Fateme Nateghi Haredasht, PhD - Stanford University; Kameron Black, DO, MPH - Stanford University; François Grolleau, MD, PhD - Stanford Center for Biomedical Informatics Research; Emily Alsentzer, MS, PhD - Brigham and Women's Hospital; Jonathan Chen, MD, PhD - Stanford University Hospital; Stephen Ma, MD, PhD - Stanford University School of Medicine;

---