A new study appearing in the Journal of the American Medical Informatics Association has made some points that may turn out to be helpful in designing those pesky but helpful alerts for clinicians.
Making alerts useful and appropriate is no small matter. As we reported on a couple of years ago, even then EMR alert fatigue has become a major source of possible medical errors. In fact, a Pediatrics study published around that time found that clinicians were ignoring or overriding many alerts in an effort to stay focused.
Despite warnings from researchers and important industry voices like The Joint Commission, little has changed since then. But the issue can’t be ignored forever, as it’s a car crash waiting to happen.
The JAMIA study may offer some help, however. While it focuses on making drug-drug interaction warnings more usable, the principles it offers can serve as a model for designing other alerts as well.
For what it’s worth, the strategies I’m about to present came from a DDI Clinical Decision Support conference attended by experts from ONC, health IT vendors, academia and healthcare organizations.
While the experts offered several recommendations applying specifically to DDI alerts, their suggestions for presenting such alerts seem to apply to a wide range of notifications available across virtually all EMRs. These suggestions include:
- Consistent use of color and visual cues: Like road signs, alerts should come in a limited and predictable variety of colors and styles, and use only color and symbols for which the meaning is clear to all clinicians.
- Consistent use of terminology and brevity: Alerts should be consistently phrased and use the same terms across platforms. They should also be presented concisely, with minimal text, allowing for larger font sizes to improve readability.
- Avoid interruptions wherever possible: Rather than freezing clinician workflow over actions already taken, save interruptive alerts that require action to proceed for the most serious situation. The system should proactively guide decisions to safer alernatives, taking away the need for interruption.
The research also offers input on where and when to display alerts.
Where to display alert information: The most critical information should be displayed on the alert’s top-level screen, with links to evidence — rather than long text — to back up the alert justification.
When to display alerts: The group concluded that alerts should be displayed at the point when a decision is being made, rather than jumping on the physician later.
The paper offers a great deal of additional information, and if you’re at all involved in addressing alerting issues or designing the alerts I strongly suggest you review the entire paper.
But even the excerpts above offer a lot to consider. If most alerts met these usability and presentation standards, they might offer more value to clinicians and greater safety to patients.
Even more important is letting the doctor take action from within the alert. If the alert suggests a certain test, why not let the MD order that test using a button in the alert?