How patients write portal messages may affect whether doctors respond

Maybe AI should write your portal messages

By Columbia University's Mailman School of Public Health

Edited by Sadie Harley, reviewed by Robert Egan

Portal messages have become a routine way for patients to communicate with their health care teams. But a new study from Columbia University Mailman School of Public Health suggests that how patients write those messages can play a key role in whether—and by whom—they receive a response. The findings are published in JAMA Network Open.

Researchers analyzed more than 3.6 million patient portal message threads and found that patients from historically marginalized groups were generally less likely to receive responses from their care teams and, in particular, were less likely to receive responses from the messages' originally intended recipient—most often their doctor. The natural next question was "why?"

While the content of a message was an important predictor of care team response, it explained little of the differences by race, ethnicity, education or insurance. Writing style—including factors such as length, formality, sentiment and how a message opened—accounted for nearly half of those differences.

"Patient portal messages are an increasingly central channel for patient-clinician communication, but there are inequities in how responsive care teams are to messages sent by patients from historically marginalized groups," said Mitchell Tang, Ph.D., assistant professor of health policy and management at Columbia Mailman School of Public Health.

"Our findings suggest that message writing style may be a potential source of bias in message triage and point to possible ways to reduce these disparities."

How triage shapes replies

Portal messaging volumes surged during the COVID-19 pandemic and remain elevated. "Unbeknownst to many patients, these messages do not usually go directly to their doctor's inbox," Tang said. In many health systems, messages go through a "triage" process in which nurses, medical assistants or other care team members review them first, determine their urgency and complexity, and decide whether to respond themselves or escalate them to the patient's doctor.

"In many message triage systems, nurses and medical assistants must quickly assess messages and make judgments about their urgency, complexity and need for escalation," Tang said. "Given those constraints, writing style may understandably influence how messages are perceived and prioritized."

In the cross-sectional study of 511,020 adult patients, the researchers analyzed 3,619,390 patient medical advice request messages between 2021 and 2023.

Disparities tied to writing style

Overall, patients from historically marginalized groups—including nonwhite patients, patients who preferred a language other than English, patients with lower educational attainment, and Medicaid or Medicare-Medicaid dual-eligible patients—were less likely to receive a response from their care team and the message's originally intended clinician. Compared with white patients, Black patients had a 3.7-percentage-point lower response rate from the intended clinician.

Using natural language processing, the researchers analyzed the content and writing characteristics of the messages and used statistical analysis to determine how much those factors contributed to differences in response among patient groups.

Writing style features accounted for 48% of the difference in intended clinician response for Black patients relative to white patients, 35% for Hispanic patients, 60.5% for patients with only a high school education compared with those with a college degree, and 43% for Medicaid patients compared with commercially insured patients.

"One of the most important writing-style features was the message's salutation," Tang said. "Simply beginning a message with 'Dear Tang' seemed to make it much more likely that Tang would end up responding."

A clue for fairer systems

The findings build on Tang's previous research in a 2024 study showing racial and ethnic disparities in the triaging of patient portal messages and 2025 research examining differences associated with insurance type and preferred language in addition to race and ethnicity.

"The results could help health systems identify potential sources of bias in portal-message triage and develop more equitable approaches to digital communication," Tang noted. "But they are also a reminder to think about how communication style can influence the way we perceive and respond to people—not only in health care, but in our everyday lives."

Publication details

Care Team Response to Patient Portal Message Content and Writing Style, JAMA Network Open (2026). DOI: 10.1001/jamanetworkopen.2026.30379

Journal information: JAMA Network Open