Are Today’s EHR Alerts Putting Clinicians, Patients at Risk?

Electronic health records (EHRs) have advanced medical practice by providing more accurate and up-to-date health information, coordinating care, and improving patient safety via automated alerts about prescriptions and patient conditions.

As the number of automated health alerts embedded in EHRs continues to rise, another EHR-related phenomenon is emerging—“alert fatigue.” Just as most Internet users overlook annoying pop-up advertisements online, physicians increasingly ignore or “override” the constant stream of health alerts.

A retrospective assessment published in the Journal of the American Medical Informatics Association in 2016, for example, found that physician overrides of allergy-related alerts rose from 83 percent in 2004 to nearly 88 percent in 2013. Other studies have found a similar override rate of between 80 to 90 percent for other types of EHR alerts. 

“The automated alert system has sort of become a victim of its own success,” said Jessica Ancker, Ph.D., an associate professor of health care policy and research at Weill Cornell Medical College. Ancker has spent her career assessing the role of information technology in medical decision making. As part of this work, she has spent years trying to determine the factors driving clinicians to overlook the alerts embedded in EHRs.

The Birth of Health Alerts

The roots of EHR alerts trace back to the late 1990s and a landmark report from the Institute of Medicine (IOM) titled “To Err Is Human.” This report grimly estimated that up to 98,000 patients die each year from medical errors, including improper prescribing practices.

Around this same time, groups began publishing data on successful outcomes following the incorporation of automated alerts into EHRs. For example, a study at Brigham and Women’s Hospital in Boston found that automated EHR warnings at the point-of-entry (when the physician inputs a prescription) led to a 50 percent drop in medical errors. The findings appeared in a 1998 paper in JAMA.

On the heels of the IOM report and these studies, many other health systems began implementing similar strategies, often known as clinical decision support systems (CDSS).

Over time, medical centers began to expand what a CDSS could do. Instead of considering only a handful of the most common drug-drug interactions, there were alerts for several hundred. In addition to interactions, physicians also received EHR notices about potential allergic reactions. Some alert programs have even incorporated “gentle reminders” to encourage best practices. “These are suggestions such as, ‘Your patient has just turned 50. Here are some topics you should discuss,’” Ancker said.

While such notices can be helpful in some cases, alert fatigue from too many notices can put both the patient and physician at risk, John Luo, M.D., a professor of psychiatry and interim chief information officer at the University of California, Riverside, told Psychiatric News. In addition to risking patient health, the information overload can reduce physician wellness and contribute to burnout, he said.

Could Reducing Alerts Cut Fatigue?

Ancker says she suspects that there are two possible factors contributing to alert fatigue. The first involves desensitization to the repetitive alert system (Ancker compares the situation to a smoke alarm that needs a new battery; the periodic reminder beep is annoying at first, but after a while the brain can tune it out). The second involves cognitive overload—the busier physicians are, the fewer alerts they will respond to.

Her research points to the cognitive overload hypothesis, because she has not found patterns suggesting that new alerts gain a good initial response, then decrease over time. If cognitive overload is to blame for clinicians overlooking EHR alerts, she suggests reducing the number of alerts could reduce fatigue.

Steven Daviss, M.D., president of the consulting group FUSE Health Strategies LLC and chair of APA’s Committee on Mental Health Information Technology (CMHIT), told Psychiatric News that efforts to reduce the volume of EHR alerts may prove challenging.

For one, hospitals may be reluctant to remove existing warnings from a system given the potential consequences, no matter how slight.

“Remember, this whole thing got started because IOM found that hundreds of thousands of people were dying due to preventable errors,” he said. But even if a health system is interested in reducing alert volume, there are questions such as how one defines “manageable” and what alerts should have priority, said Daviss, who is also the APA Assembly recorder. Physicians, nurses, and administrators likely have different answers to those questions, he added.

EHR systems can be customized to meet the needs of an organization, so specialty health systems or private practices can work to create an appropriate level of alerts. Daviss noted companies like Valant make EHRs tailored to mental health care, and this specificity allows them to include only psychiatry-relevant alerts (Daviss has no financial relationships to Valant). For interested parties, Daviss pointed out the CMHIT has an online resource on how to choose the right EHR for a practice.

Large and diverse health systems “must be more generic,” he said, to maintain consistency from an administrative and billing standpoint. But a generic system is a contributor to fatigue since not every alert is relevant to every physician. “For someone not versed in mental health, an alert about the risk of mixing fluoxetine with tramadol [possibly causing high serotonin levels that could lead to serotonin syndrome] is valuable, but I am aware of that interaction and do not need it repeated every time I prescribe Prozac,” Daviss said.

Some third-party groups have been working to make EHRs more flexible; a company called First Databank has developed a CDSS tool called AlertSpace that enables hospitals to customize which alerts are enabled or disabled in different departments, and some health care systems have reported some success with these platforms.

Though options are available, Daviss believes that research is critical to find the right balance of alerts per patient, just as automated CDSSs were first developed from evidence collected from controlled clinical studies. “It’s much like other technologies; we do need rigorous testing to reach a level of consistent quality.”

In the meantime, physicians should try other strategies to avoid getting frustrated by the pervasive health alert, Luo told Psychiatric News. “These alerts are just one aspect of our digital, connected world,” he said. There may not be much one can do about an EHR pop-up, but he suggests physicians turn off their phones or other devices during their down time to avoid compounding this information overload. ■