EPtalk by Dr. Jayne 1/8/26

Mr. H’s mention of a recent article caught my attention. It says that 40 million people are using ChatGPT for health-related questions every day.

I agree with the article’s statement that people are “turning to AI tools to navigate the notoriously complex and opaque US healthcare system.” They mention patients using it to decipher billing statements, appeal insurance denials, and answer clinical questions “when access to doctors is limited.”

Another statistic that caught my attention: more than 5% of ChatGPT questions are about healthcare, and 1.6 million questions per week are asked about health insurance.

Clinicians certainly can’t fault patients for using AI tools when they are doing the same. I see physicians every day using AI to write insurance appeals and create patient-facing communications, not to mention all the AI-powered documentation. The risk of hallucinations remains a major concern. Some care delivery organizations have applied their “we can’t control it so we’ll just ignore it” philosophy. 

I would instead encourage organizations to make better use of their existing tools in providing accurate and vetted information to patients. Those institutions that offer robust patient education and engagement solutions should feature that information prominently on their websites and within their patient portals. Patients would be able to self-serve with reputable information.

Clinicians need to look at patient education less as a check-the-box exercise and more as a key part of patient care. In my experience, educated patients who have access to resources that they can consult down the road are less likely to send patient portal messages or call the office with basic questions. They feel more confident about their care and their ability to manage at home.

Another juicy tidbit from the report: 70% of health-related ChatGPT queries occur outside of normal medical office hours. Most medical offices are open for about eight hours per day, usually overlapping the same work hours as people who also work traditional schedules. It’s difficult for many patients and caregivers to get the information they need during the hours that they are available. Patient portals and secure messaging have helped this issue somewhat, but gaps still exist.

In addition to making sure that patients know how to access trustworthy patient education materials, care delivery organizations should do a better job promoting other patient-facing resources, such as after-hours nurse triage lines or on-call services. Organizations that are actively managing risk do a better job with this, because they are incentivized to keep patients from going to the emergency department.

It would be interesting to compare after-hours use of generative AI solutions by patients who have access to after-hours services and those who don’t. Anyone up for some research?

From Midwest Gal: “Re: portal messages. You mentioned waiting for test results, received a patient portal notification that you had a message from the physician, and it turned out it was a general message about holiday hours. The same thing happened to me right before the Christmas holiday. Instead of getting my mammogram results, it was a reminder that the office would be closed.” I reached out to some folks who are experts in the EHR that the reader’s site uses. They said that using the patient portal in this manner is not a best practice. For the love of all things, if you’re on a patient portal team, please work with the operations teams that are sending these messages to help them understand the anxiety that they are causing.

Speaking of anxiety, the clinical trial in which I am a participant published some of its results recently. However, it didn’t bother to notify patients that this would be happening. Those of us that are clinicians saw it in the journals first, which was bad enough. To make things worse, the research team released new recommendations to patients several days later, some of which provided guidance that is counter to the standard of care. That was accompanied by no explanation.

This occurred the week of December 18, when many people are frazzled by year-end work responsibilities or holiday preparations. I can’t imagine a worse time to release that kind of information.

I reached out to the study coordinator with my questions. I didn’t receive a reply within the published service level, so I reached out again via a different method. Guess what? They were experiencing a high volume of calls and were short staffed due to the holidays. The local physician who had referred me to the study wasn’t aware of either the published article or the communication to patients. You really cannot make this stuff up.

From Burned Out CMIO: “Re: help desk. My large health system outsourced its help desk functions at the beginning of December with the assurance that we would see no degradation in service levels. I had complaints from my ED physicians, who said that their tickets had been closed due to lack of customer response. Help desk staff were emailing the physicians about their tickets, then closing them as unresponsive if they didn’t hear back within a few hours. We’ve been having some serious conversations with the vendor about how that’s not how it’s supposed to work, especially for shift-based physicians who might not be able to respond quickly and then might not be working the next day. Ambulatory physicians ran into issues during Christmas week when offices were closed some days, then came back on Monday to find their tickets closed due to ‘no response from customer.’ Everything blew up over the New Year’s holiday, when tickets were closed in bulk on the 31st to meet meet end-of-year service level metrics. I feel awful because people who I had worked with for years were laid off in favor of the allegedly cheaper outsource firm.”

In situations like this, you can’t put a price on the knowledge of former help desk staffers who understood user and office work schedules around the holidays. I wonder if this outsource firm has any healthcare experience. This falls into the category of “you get what you paid for.”

I hope that a robust review of service level expectations happens again and that ticket closure goals are moved out a bit to accommodate the behaviors of real users in the healthcare setting. I can just imagine people trying to slam tickets shut to meet the metrics, not realizing that users have valid reasons for not responding quickly.

What’s the most foolish outsource maneuver your organization has made? Leave a comment or email me.

Email Dr. Jayne.