Patient portal message responses vary by race, ethnicity

The doctor is in? For Black patients sending patient portal messages, that might not be true, according to a new study in JAMA Network Open.

The analysis of nearly 58,000 patient portal messages sent by patients at Boston Medical Center showed that although Black and other ethnic minority patients were just as likely to receive a patient portal message response, they were less likely to get one from an attending physician.

This trend could help perpetuate racial health disparities, the researchers reported.

Patient portal messaging became significantly more common during the COVID-19 pandemic. For one thing, the pandemic prompted more patients and their providers to leverage health IT to connect remotely when possible. Additionally, COVID-19 introduced a litany of health-related questions and needs for which patients queried providers using their patient portal accounts.

Previous studies have found that certain patient populations are more likely to use the patient portal than others, which active users tending to skew Whiter and younger. In January 2023, the ONC found that more White patients used the patient portal largely because they were more likely to be invited by their providers to use the technology.

This latest study gives a further look into health IT disparities, showing that even when racial/ethnic minority patients use technology, their experiences might not be the same as their White counterparts.

The researchers, who hailed from BMC, Harvard, and Mass General Brigham, included nearly 40,000 patients of White, Black, Hispanic, and Asian race in the study to assess the 57,704 patient portal messages they sent between January and November 2021.

Overall, patients who belonged to minoritized racial or ethnic groups were just as likely to get their patient portal message answered compared to their White counterparts. However, they were less likely to get an answer from an attending physician.

For example, Black patients were 3.95 percentage points less likely to hear back from an attending physician compared to White patients. Instead, they were 3.01 percentage points more likely to hear back from a registered nurse.

Similar but smaller trends emerged for Asian and Hispanic patients. Asian patients were 2.11 percentage points less likely to hear back from an attending physician. For Hispanic patients, that figure was 2.32 percentage points.

It’s worth noting that many nurses are tasked with answering patient portal messages, and they provide high-quality remote care. However, the disparities still warrant a closer look, the researchers said.

“The central tenet behind message triaging is that physician time is valuable and should be allocated judiciously,” they explained. “In our setting, physician resources—as measured by threads responded to—were disproportionately allocated toward White patients. While only 21.1% of patients in our sample were White, they received 46.3% of all attending physician responses.”

The difference in patient portal message response is likely due to how each message gets forwarded to hospital staff, the researchers said. At BMC, patient portal messages get pooled and then forwarded by triage nurses. An attending physician cannot answer a portal message that isn’t sent to them.

Differences in who the triage nurse forwards a patient portal message to might be explained by the underlying message request or even the way the message was communicated. There is also the high risk of implicit bias coming into play, the research team added.

Moving forward, researchers should look at the reasons behind these differences in patient portal message response. Doing so is critical, as certain explanations may be a bigger red flag than others.

For example, differences in patient requests could indicate a practice needs to tailor patient education. Improving patient portal health literacy could help certain populations better understand the best uses for the technology.

Differences in patient portal response that are due to implicit bias should prompt organizations to reexamine their implicit bias training to stress not just in-person interactions but also the way grammar, sentence structure, and tone can shape perceptions.

Importantly, the researchers stressed that it is not necessarily a bad thing for a registered nurse to be the primary correspondent on a patient portal message, noting that they aren’t aware of any benefits of an attending physician versus a nurse responding to messages.

“Indeed, for many types of message requests, direct RN response may be most appropriate,” they concluded. “However, when differences in triage prioritization are driven not by differences in the underlying message request itself, but rather by differences in the way the message is communicated or by (potentially unconscious) care team bias, we should be particularly concerned about disparities in care team resource allocation and consider potential levers to mitigate those inequities.”